CN112101804A - Vehicle scheduling method and device, readable storage medium and electronic equipment - Google Patents
Vehicle scheduling method and device, readable storage medium and electronic equipment Download PDFInfo
- Publication number
- CN112101804A CN112101804A CN202010997098.8A CN202010997098A CN112101804A CN 112101804 A CN112101804 A CN 112101804A CN 202010997098 A CN202010997098 A CN 202010997098A CN 112101804 A CN112101804 A CN 112101804A
- Authority
- CN
- China
- Prior art keywords
- time slice
- attribute
- sequence
- vector
- graph
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 239000013598 vector Substances 0.000 claims abstract description 171
- 230000000737 periodic effect Effects 0.000 claims abstract description 17
- 230000007246 mechanism Effects 0.000 claims abstract description 15
- 238000010606 normalization Methods 0.000 claims abstract description 13
- 230000006870 function Effects 0.000 claims description 13
- 238000004590 computer program Methods 0.000 claims description 12
- 238000007477 logistic regression Methods 0.000 claims description 3
- 230000008859 change Effects 0.000 abstract description 6
- 238000010586 diagram Methods 0.000 description 14
- 230000008569 process Effects 0.000 description 5
- 230000004048 modification Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000002093 peripheral effect Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000011084 recovery Methods 0.000 description 1
- 239000002699 waste material Substances 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/26—Government or public services
Landscapes
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Engineering & Computer Science (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Tourism & Hospitality (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- General Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Physics & Mathematics (AREA)
- Quality & Reliability (AREA)
- Operations Research (AREA)
- Game Theory and Decision Science (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The embodiment of the invention discloses a vehicle scheduling method, a vehicle scheduling device, a readable storage medium and electronic equipment. The method comprises the steps of determining a target area comprising a plurality of parking areas, determining an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, inputting a plurality of graph convolution layers respectively, and outputting corresponding graph characteristics. And inputting the time slice sequence and the characteristics of each graph into an attention mechanism layer to determine corresponding attribute characteristic vectors, and inputting the time slice sequence and the characteristics of each graph into a full connection layer after normalization processing to determine attribute value ranges corresponding to the target time slices of each parking area for vehicle scheduling. According to the embodiment of the invention, the chart convolution layers are respectively trained through different parameters such as the distance between parking areas in the target area, the number of facilities around the parking areas, road network information of the parking areas and the like, so that the attribute change of the parking areas is comprehensively evaluated from different angles, a predicted attribute value range is obtained, and the accuracy and reliability of attribute value prediction are improved.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a vehicle scheduling method, a vehicle scheduling device, a readable storage medium and electronic equipment.
Background
At present, the shared electric bicycle is widely applied to various big cities, and great convenience is brought to the life of people. In order to facilitate the management of the shared electric bicycle and the use of users, the dispatching platform is provided with a plurality of parking spots. The supply amount of vehicles and the demand of users for vehicles at different times are different for each parking spot, and are interfered by the parking spot position, environment, temporary scheduling of vehicles, and the like. In order to ensure that the vehicle supply amount of each parking point can meet the vehicle demand amount, the supply amount and the demand amount of each parking point at the target time need to be predicted in advance, so that the vehicle can be dispatched according to the predicted value.
Disclosure of Invention
In view of this, embodiments of the present invention provide a vehicle scheduling method, an apparatus, a readable storage medium, and an electronic device, which are used to comprehensively evaluate attribute changes of a parking area from different angles to obtain a predicted attribute value range, and schedule a vehicle according to the predicted value.
In a first aspect, an embodiment of the present invention provides a vehicle scheduling method, where the method includes:
determining a target area, wherein the target area comprises a plurality of parking areas;
determining an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, wherein the attribute vector sequence comprises a plurality of attribute vectors corresponding to each time slice in the time slice sequence, the attribute vectors comprise attribute values corresponding to each parking area, and the attribute values are used for representing vehicle demand or vehicle supply of the corresponding parking areas;
inputting the attribute vector sequence into a plurality of graph convolution layers respectively so as to output a plurality of corresponding graph characteristics respectively;
determining an attribute value range corresponding to each parking area in a target time slice according to the time slice sequence and each graph characteristic, wherein the interval duration between the target time slice and the last time slice in the time slice sequence is one period of the time slice sequence;
and carrying out vehicle dispatching on each parking area according to the corresponding attribute value range.
In a second aspect, an embodiment of the present invention provides a vehicle dispatching device, where the device includes:
the parking system comprises an area determining module, a parking area determining module and a parking area determining module, wherein the area determining module is used for determining a target area, and the target area comprises a plurality of parking areas;
the attribute determining module is used for determining an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, wherein the attribute vector sequence comprises a plurality of attribute vectors corresponding to each time slice in the time slice sequence, the attribute vectors comprise attribute values corresponding to each parking area, and the attribute values are used for representing the vehicle demand or the vehicle supply quantity of the corresponding parking area;
the graph characteristic determining module is used for respectively inputting the attribute vector sequence into a plurality of graph convolution layers so as to respectively output a plurality of corresponding graph characteristics;
the prediction module is used for determining an attribute value range corresponding to each parking area in a target time slice according to the time slice sequence and each graph characteristic, and the interval duration between the target time slice and the last time slice in the time slice sequence is one period of the time slice sequence;
and the scheduling module is used for scheduling the vehicles in each parking area according to the corresponding attribute value range.
In a third aspect, the present invention provides a computer-readable storage medium for storing computer program instructions, which when executed by a processor implement the method according to any one of the first aspect.
In a fourth aspect, an embodiment of the present invention provides an electronic device, including a memory and a processor, the memory being configured to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method according to any one of the first aspect.
According to the embodiment of the invention, the target area comprising a plurality of parking areas is determined, the attribute vector sequence corresponding to the target area is determined according to the preset periodic time slice sequence, then the plurality of graph convolution layers are respectively input, and the corresponding graph characteristics are output. And inputting the time slice sequence and the characteristics of each graph into an attention mechanism layer to determine corresponding attribute characteristic vectors, and inputting the time slice sequence and the characteristics of each graph into a full connection layer after normalization processing to determine attribute value ranges corresponding to the target time slices of each parking area for vehicle scheduling. According to the embodiment of the invention, the chart convolution layers are respectively trained through different parameters such as the distance between the parking areas in the target area, the number of facilities around the parking areas, road network information of the parking areas and the like, so that the attribute change of the parking areas is comprehensively evaluated from different angles, and the accuracy and reliability of attribute value prediction are improved. And finally obtaining a predicted attribute value range, namely the vehicle supply amount range and the vehicle demand amount range of each parking area, and scheduling the vehicles according to the predicted values.
Drawings
The above and other objects, features and advantages of the present invention will become more apparent from the following description of the embodiments of the present invention with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of a vehicle scheduling method according to an embodiment of the present invention;
FIG. 2 is a schematic illustration of a target area according to an embodiment of the present invention;
FIG. 3 is a schematic view of a parking area according to an embodiment of the present invention;
FIG. 4 is a diagram illustrating a Euclidean distance determination for a parking area according to an embodiment of the present invention;
FIG. 5 is a diagram illustrating an embodiment of determining the number of information points around a parking area;
FIG. 6 is a schematic diagram illustrating an embodiment of determining road information around a parking area;
FIG. 7 is a diagram illustrating an exemplary method for determining attribute value ranges corresponding to parking areas according to an embodiment of the present invention;
FIG. 8 is a schematic diagram of a vehicle dispatching device according to an embodiment of the present invention;
fig. 9 is a schematic diagram of an electronic device according to an embodiment of the invention.
Detailed Description
The present invention will be described below based on examples, but the present invention is not limited to only these examples. In the following detailed description of the present invention, certain specific details are set forth. It will be apparent to one skilled in the art that the present invention may be practiced without these specific details. Well-known methods, procedures, components and circuits have not been described in detail so as not to obscure the present invention.
Further, those of ordinary skill in the art will appreciate that the drawings provided herein are for illustrative purposes and are not necessarily drawn to scale.
Unless the context clearly requires otherwise, throughout the description, the words "comprise", "comprising", and the like are to be construed in an inclusive sense as opposed to an exclusive or exhaustive sense; that is, what is meant is "including, but not limited to".
In the description of the present invention, it is to be understood that the terms "first," "second," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance. In addition, in the description of the present invention, "a plurality" means two or more unless otherwise specified.
Fig. 1 is a flowchart of a vehicle scheduling method according to an embodiment of the present invention, and as shown in fig. 1, the vehicle scheduling method includes the following steps:
and step S100, determining a target area.
Specifically, the target area may be directly selected from a plurality of areas pre-stored in the server, or may be temporarily determined according to information such as a received user instruction. The target area comprises a plurality of parking areas, and the parking area is a specific sub-area in the target area. For example, in a shared electric bicycle platform, the target area may be a preset specific operation area such as a business district, a cell, a playground, a scenic spot and the like in the shared electric bicycle operation city, and the parking area is a preset parking area in the target area.
Fig. 2 is a schematic diagram of a target area according to an embodiment of the present invention. As shown in fig. 2, the target area 20 is a preset area, and includes a plurality of parking areas 21. Each of the parking areas 21 is used for parking a shared electric bicycle, and the supply amount and the demand amount of the vehicle change with the change of vehicle scheduling, recovery, and the like.
Fig. 3 is a schematic view of a parking area according to an embodiment of the present invention. As shown in fig. 3, in the shared electric bicycle platform, the parking area 30 is a parking area in which a plurality of shared electric bicycles 31 are parked. And the user of the shared electric bicycle platform takes and parks the shared electric bicycle based on the parking area.
And S200, determining an attribute vector sequence corresponding to the target region according to a preset periodic time slice sequence.
Specifically, the period, the time slice length, and the sequence length of the time slice sequence may be preset. Wherein the period may be, for example, a day, a week, a month, etc., and the time slice length may be, for example, a half hour, an hour, a day, etc. The attribute vector sequence comprises a plurality of attribute vectors corresponding to each time slice in the time slice sequence, and each attribute vector represents the attribute of the target region in the corresponding time slice. And each element in the attribute vector is an attribute value corresponding to each parking area and is used for representing the vehicle demand or the vehicle supply of the corresponding parking area. In the field of electric bicycle sharing, the vehicle supply amount changes due to reasons such as increasing the number of parking areas, decreasing the number of parking areas, and changing, collecting, and scheduling the vehicles in the parking areas, which are located in the vicinity of the parking areas. Thus, the attribute vectors corresponding to different time slices in the sequence of attribute vectors are different.
In an optional implementation manner of the embodiment of the present invention, the attribute value is a vehicle demand, and is used to characterize a vehicle demand situation of a user in a corresponding time slice in a corresponding parking area. Therefore, the process of determining the attribute vector sequence corresponding to the target region includes the following steps:
step S210, respectively obtaining the number of vehicles in each parking area in the target area at the starting time point and the ending time point of each time slice in a preset periodic time slice sequence.
Specifically, since the time slice is a time period, when determining the attribute vector corresponding to each time slice in the periodic time slice sequence, the attribute values of the parking areas in the target area need to be determined at the starting time point and the ending time point of each time slice respectively. The vehicle scheduling method is applied to a shared electric bicycle platform, and the time slices included in the time slice sequence are respectively 17:00-18:00 per day in the previous week. When determining the historical demand condition of vehicles in each parking area in the target area, the number of the vehicles parked in each parking area at 17:00 pm and the number of the vehicles parked in each parking area at 18:00 pm in the previous week are obtained.
Step S220, for each parking area, determining a corresponding vehicle demand according to the number of vehicles at the starting time point and the number of vehicles at the ending time point of each time slice.
Specifically, the vehicle demand amount corresponding to each parking area in different time slices may be determined according to the number of vehicles at the starting time point and the number of vehicles at the ending time point. Optionally, the vehicle demand amount is a difference between the number of vehicles at the start time point and the number of vehicles at the end time point. For example, when the number of vehicles in a parking area at the start time point of a corresponding time slot is 20 and the number of vehicles at the end time point is 10, it is determined that the vehicle demand of the parking area at the time slot is 10.
Step S230, determining an attribute vector sequence according to the vehicle demand corresponding to each parking area in each time slice in the target area.
Specifically, after the vehicle demand of each parking area in the target area in each time slice is determined, the corresponding attribute vector is determined according to the vehicle demand of each parking area in each time slice, and then the attribute vector sequence is determined according to the sequence of each time slice in the time slice sequence.
In another optional implementation manner of the embodiment of the present invention, the attribute value is a vehicle supply amount, and is used for characterizing vehicle supply conditions of users in a corresponding parking area in a corresponding time slice. Therefore, the process of determining the attribute vector sequence corresponding to the target region includes the following steps:
step S210', obtaining the number of vehicles in each parking area in the target area at the starting time point of each time slice in a preset periodic time slice sequence.
In the embodiment of the present invention, the process of determining the number of vehicles in each target area at the starting time point of each time slice is similar to step S210, and is not described herein again.
In step S220', for each parking area, the corresponding vehicle supply amount is determined based on the number of vehicles at each time slice start time.
Specifically, the vehicle supply amount corresponding to each of the parking areas at different time slices may be directly determined as the number of vehicles at the start time point. For example, when the number of vehicles in one parking area at the corresponding time slot start time point is 20, it is determined that the vehicle supply amount in the parking area at the time slot is 20.
Step S230' determines an attribute vector sequence according to the vehicle supply amount corresponding to each time slice of each parking area in the target area.
Specifically, after the vehicle supply amount of each parking area in the target area in each time slice is determined, the corresponding attribute vector is determined according to the vehicle supply amount of each parking area in each time slice, and then the attribute vector sequence is determined according to the sequence of each time slice in the time slice sequence.
Step S300, inputting the attribute vector sequence into a plurality of graph convolution layers respectively so as to output a plurality of corresponding graph features respectively.
Specifically, after the attribute vector sequence is determined, the attribute vector sequence is respectively input into a plurality of graph convolution layers obtained by training according to different training sets to determine a plurality of corresponding graph features. In the embodiment of the invention, the attribute vector sequence is respectively input into a first graph convolutional layer, a second graph convolutional layer and a third graph convolutional layer which are obtained by pre-training so as to output corresponding first graph characteristics, second graph characteristics and third graph characteristics. And each convolutional layer is obtained by training according to a plurality of feature map sets influencing the vehicle supply and demand conditions in the parking area.
The vehicle scheduling method is applied to a shared electric bicycle platform as an example for explanation. The supply and demand conditions of several parking areas in the same target area are affected by adjacent parking areas, peripheral information points, and road network information. For example, when the vehicle supply of one parking area fails to meet the user's demand, the vehicle demand of a nearby parking area that can provide a sufficient supply will rise. The peak value of the vehicle demand of the parking area near the scenic spot is from friday night to sunday night, and the vehicle demand is little in the working day period. The peak value of the vehicle demand in the parking area near the office building is generally at the working peak of work on and off duty, the demand of the working peak is concentrated, and the demand of the working peak of work off duty is divergent. Meanwhile, the vehicle demand amount is generally higher in a parking area where there are many nearby roads than in a parking area where there are few nearby roads. Therefore, the first map convolutional layer, the second map convolutional layer and the third map convolutional layer can be obtained by training according to the vector set corresponding to the information influencing the vehicle supply and demand condition.
Specifically, the first graph convolution layer is determined by training according to a set of euclidean distance vectors corresponding to the target region, where the set of euclidean distance vectors includes a plurality of euclidean distance vectors. Each Euclidean distance vector is determined according to the distance between every two parking areas in the target area, and each element in the Euclidean distance vectors corresponds to one parking area and comprises the distance between the parking area and other parking areas in the target area.
Fig. 4 is a schematic diagram of determining the euclidean distance of the parking area according to the embodiment of the present invention. As shown in fig. 4, the distance 40 between two parking areas in the target area is determined as follows. Firstly, respectively determining the longitude and latitude coordinates coor of a parking area P1 and a parking area P2 in a target area1And coor2Then according to the longitude and latitude coordinates coor1And coor2Calculate the Euclidean distance (color) between the parking areas P1 and P21,coor2)。
And the second graph convolution layer is determined according to the region information vector set corresponding to the target region, wherein the region information vector set comprises a plurality of region information vectors. Each area information vector is determined according to the number of information points around every two parking areas in the target area, each element in the area information vector corresponds to one parking area, and the information points around the parking areas and other parking areas in the target area are determined according to the number of information points.
Fig. 5 is a schematic diagram illustrating the determination of the number of information points around the parking area according to the embodiment of the invention. As shown in fig. 5, one element of the region information vector is determined as follows. Firstly, the number poi of information points in the preset range around the parking area P1 and the parking area P2 in the target area is determined1And poi2. The information points may be preset, and may include, for example, banks, hospitals, schools, supermarkets, gourmets, scenic spots, hotels, and the like, that is, poi1And poi2The number of information points in the parking area P1 and the number of information points in the parking area P2, such as banks, hospitals, schools, supermarkets, etc., are included. Further according to the number poi of the peripheral multiple information points1And poi2Calculating cosine similarity cosine (poi)1,poi2) The corresponding element is obtained to determine the region information vector.
And the third graph convolution layer is determined according to the training of a road network information vector set corresponding to the target area, wherein the road network information vector set comprises a plurality of road network information vectors. The road network information vectors are determined according to road information around every two parking areas in the target area, each element in the road network information vector corresponds to one parking area, and the road network information vectors are determined according to the road network information around the parking areas and other parking areas in the target area.
Fig. 6 is a schematic diagram illustrating a determination of road information around a parking area according to an embodiment of the present invention. As shown in fig. 6, one element in the road network information vector is determined as follows. Firstly, determining road network information road _ net in preset ranges around a parking area P1 and a parking area P2 in the target area1And road _ net2. The road network information may include a plurality of preset road network characteristics, such as intersection density, road density, and averageConnectivity, mean penetration, centrality, etc. Namely, road _ net1And road _ net2Respectively, for characterizing road conditions around the parking area P1 and the parking area P2. And then further according to the condition of the surrounding road, road _ net1And road _ net2Calculating the pre-similarity cosine (road _ net)i,road_netj) And obtaining corresponding elements to determine the road network information vector.
And step S400, determining the attribute value range corresponding to the parking areas in the target time slice according to the time slice sequence and the image characteristics.
Specifically, the time duration of the interval between the target time slice and the last time slice in the time slice sequence is one period of the time slice sequence. For example, when the time slices included in the time slice sequence are respectively 17:00-18:00 per day of the previous week, the target time slice may be 17:00-18:00 per day. In this embodiment of the present invention, the process of determining the attribute value range corresponding to the target time slice in each parking area includes the following steps:
and S410, inputting the time slice sequence and the graph characteristics into an attention mechanism layer to determine corresponding attribute characteristic vectors.
Specifically, the attention mechanism layer in the embodiment of the present invention is a Multi-headed attention mechanism layer (Multi-headed attention mechanism). And after the time slice sequence is input into the attention mechanism layer, performing point multiplication on the time slice sequence and a preset position code, and combining the time slice sequence and each graph feature to obtain a corresponding attention input vector. The attention input vector determines a plurality of corresponding attention feature vectors according to a plurality of attention weight matrixes in the attention mechanism layer, and calculates each attention feature vector logistic regression function (softmax) to obtain an attribute feature vector.
Step S420, after normalization processing of the attribute feature vectors, inputting the normalized attribute feature vectors into a full connection layer to determine an attribute value range corresponding to each parking area in a target time slice.
Specifically, after the attribute feature vector is determined by the attention mechanism layer, the attribute feature vector needs to be normalized. In an embodiment of the present invention, the normalizing includes inputting the attribute feature vector into a first normalization layer, for normalizing values of all dimensions of a sample with a single sample as a target, and outputting a first feature vector. And inputting the first eigenvector into a front feedback layer for linear conversion, and outputting a second eigenvector. And finally, merging the first feature vector and the second feature vector, inputting the merged first feature vector and second feature vector into a second normalization layer, and normalizing again to obtain a third feature vector. Further, after the attribute feature vector is normalized to obtain a third feature vector, the third feature vector is input into a full-connection layer, so that the attribute value range of each parking area corresponding to the target time slice is determined through the loss function corresponding to the full-connection layer. The process of determining the attribute value range corresponding to the target time slice of each parking area comprises the following steps:
step S421, inputting the third feature vector into a full-connected layer, and calculating and determining a plurality of predicted values corresponding to each parking area according to a quantile regression function and a plurality of preset quantiles.
Specifically, in order to increase the reliability of the final predicted value, the loss function in the embodiment of the present invention is a quantile regression function, which includes a plurality of preset quantiles, so as to output the predicted values corresponding to the preset quantiles respectively. For example, when the preset quantiles are 25%, 50% and 75%, respectively, the third feature vector is input to the fully-connected layer, and then three predicted values corresponding to 25%, 50% and 75% of each parking area are output.
Step S422, determining an attribute value range corresponding to the target time slice in each parking area according to a preset output rule and the corresponding plurality of predicted values.
Specifically, in the embodiment of the present invention, the output rule may be determined according to an attribute value determined as a prediction from one of the plurality of prediction values, and then determined according to another prediction value to determine the attribute value range. For example, when the preset quantiles are 25%, 50% and 75%, respectively, the output rule may be to determine the reliable section according to the predicted values corresponding to 25% and 75%, and determine the attribute value range by using the predicted value corresponding to 50% as the predicted attribute value. And determining the reliable interval according to the difference value of the predicted values corresponding to 25% and 75%. When the predicted values corresponding to 25%, 50% and 75% of the preset quantiles are 10, 15 and 20 respectively, the attribute value range is 15 +/-10.
And step S500, vehicle scheduling is carried out on each parking area according to the corresponding attribute value range.
Specifically, after the attribute value range corresponding to each parking area is determined, vehicles are scheduled to the parking areas according to the corresponding attribute value range, so that the vehicle supply quantity of each parking area can meet the vehicle demand quantity, and meanwhile, waste of vehicle resources is avoided. For example, when it is predicted that the vehicle demand amount range corresponding to a parking area in a target time slot is 15 ± 10 and the corresponding vehicle supply amount range is 5 ± 10, respectively, it is necessary to schedule 10 electric bicycles like the parking area before the target time slot arrives.
Fig. 7 is a schematic diagram illustrating the determination of the attribute value range corresponding to each parking area according to the embodiment of the present invention. As shown in fig. 7, an attribute vector sequence 70 corresponding to a target region is determined by a preset time slice sequence 74, and the attribute vector sequence 70 is input into a first graph convolution layer 71, a second graph convolution layer 72 and a third graph convolution layer 73, respectively, so as to output corresponding first graph feature, second graph feature and third graph feature, respectively. The time slice sequence 74, the first graph feature, the second graph feature, and the third graph feature are input to the attention mechanism layer 75, and corresponding attribute feature vectors are output. The attribute feature vector is normalized by a first normalization layer 76 to obtain a first feature vector, and the first feature vector is input to a front feedback layer 77 to be linearly converted to obtain a second feature vector. The first feature vector and the second feature vector are input to the second normalization layer 78, and a third feature vector is output. The third feature vector passes through the full connection layer 79, so that a plurality of predicted values corresponding to each parking area are calculated and determined according to a quantile regression function and a plurality of preset quantiles, and an attribute value range 7A corresponding to each parking area is further determined.
According to the vehicle scheduling method, the chart convolution layers are trained respectively through different parameters such as the distance between parking areas in the target area, the number of facilities around the parking areas, road network information of the parking areas and the like, so that the attribute change of the parking areas is comprehensively evaluated from different angles, a predicted attribute value range is obtained, and the accuracy and reliability of attribute value prediction are improved.
Fig. 8 is a schematic diagram of a vehicle scheduling apparatus according to an embodiment of the present invention, as shown in fig. 8, the vehicle scheduling apparatus includes a region determining module 80, an attribute determining module 81, a map feature determining module 82, a predicting module 83, and a scheduling module 84.
Specifically, the area determination module 80 is configured to determine a target area, which includes a plurality of parking areas. The attribute determining module 81 is configured to determine an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, where the attribute vector sequence includes a plurality of attribute vectors corresponding to each time slice in the time slice sequence, the attribute vectors include attribute values corresponding to each parking area, and the attribute values are used to represent vehicle demand or vehicle supply of the corresponding parking area. The graph feature determination module 82 is configured to input the attribute vector sequence into a plurality of graph convolution layers, respectively, so as to output a plurality of corresponding graph features, respectively. The prediction module 83 is configured to determine, according to the time slice sequence and the map features, an attribute value range corresponding to each parking area in a target time slice, where an interval duration between the target time slice and a last time slice in the time slice sequence is one period of the time slice sequence. The scheduling module 84 is configured to schedule vehicles in each of the parking areas according to the corresponding attribute value range.
Further, the attribute value is a vehicle demand;
the attribute determination module includes:
the first vehicle number determining unit is used for respectively acquiring the vehicle number of each parking area in the target area at the starting time point and the ending time point of each time slice in a preset periodic time slice sequence;
a vehicle demand determination unit for determining a corresponding vehicle demand amount for each of the parking areas based on the number of vehicles at the start time point and the number of vehicles at the end time point of each of the time slices;
and the first attribute determining unit is used for determining an attribute vector sequence according to the vehicle demand corresponding to each parking area in the target area in each time slice.
Further, the attribute value is a vehicle supply amount;
the attribute determination module includes:
a second vehicle number determination unit, configured to obtain the number of vehicles in each parking area in the target area at a start time point of each time slice in a preset periodic time slice sequence;
a vehicle supply determination unit configured to determine, for each of the parking areas, a corresponding vehicle supply amount based on the number of vehicles at each of the time slice start time points;
and the second attribute determining unit is used for determining an attribute vector sequence according to the vehicle supply quantity corresponding to each parking area in each time slice in the target area.
Further, the graph feature determination module specifically includes:
and the graph characteristic determining unit is used for inputting the attribute vector sequence into the first graph convolutional layer, the second graph convolutional layer and the third graph convolutional layer respectively so as to output corresponding first graph characteristic, second graph characteristic and third graph characteristic.
Further, the first map convolutional layer is determined according to Euclidean distance vector set training corresponding to the target region, and each Euclidean distance vector in the Euclidean distance vector set is determined according to the distance between every two parking regions in the target region.
Further, the second map convolution layer is determined by training according to an area information vector set corresponding to the target area, and each area information vector in the area information vector set is determined according to the number of information points around each two parking areas in the target area.
Further, the third graph convolution layer is determined according to a road network information vector set corresponding to the target region, and each road network information vector in the road network information vector set is determined according to road information around every two parking regions in the target region.
Further, the prediction module comprises:
a feature vector determination unit, configured to input the time slice sequence and each of the graph features into an attention mechanism layer to determine a corresponding attribute feature vector;
and the prediction unit is used for performing normalization processing on the attribute feature vectors and inputting the normalized attribute feature vectors into a full-link layer so as to determine the attribute value range corresponding to each parking area in a target time slice.
Further, the feature vector determination unit includes:
a first vector determining subunit, configured to perform point multiplication on the time slice sequence and a preset position code, and then combine the time slice sequence with each of the graph features to determine a corresponding attention input vector;
a second vector determination subunit for determining a corresponding attention feature vector from the plurality of attention weight matrices in the attention mechanism layer and an attention input vector;
and the characteristic vector determining unit is used for determining the attribute characteristic vector according to each attention characteristic vector and the logistic regression function.
Further, the prediction unit includes:
a third vector determination subunit, configured to input the attribute feature vector into the first normalization layer to determine a first feature vector;
a fourth vector determination subunit for inputting the first eigenvector into a feedforward layer to determine a second eigenvector;
a fifth vector determining subunit, configured to input the first feature vector and the second feature vector into a second normalization layer to determine a third feature vector;
and the predicting subunit is used for inputting the third feature vector into a full-link layer so as to determine an attribute value range corresponding to each parking area in a target time slice.
Further, the predictor unit includes:
the first prediction subunit is used for inputting the third feature vector into a full-connection layer so as to calculate and determine a plurality of predicted values corresponding to the parking areas according to a quantile regression function and a plurality of preset quantiles;
and the second prediction subunit is used for determining the attribute value range corresponding to each parking area in the target time slice according to a preset output rule and a plurality of corresponding prediction values.
The vehicle dispatching device provided by the embodiment of the invention respectively trains the chart convolution layers through different parameters such as the distance between parking areas in the target area, the number of facilities around the parking areas, road network information of the parking areas and the like, so that the attribute change of the parking areas is comprehensively evaluated from different angles, a predicted attribute value range is obtained, and the accuracy and reliability of attribute value prediction are improved.
Fig. 9 is a schematic diagram of an electronic device according to an embodiment of the invention. As shown in fig. 9, the electronic device shown in fig. 9 is a general address query device, which includes a general computer hardware structure, which includes at least a processor 90 and a memory 91. The processor 90 and the memory 91 are connected by a bus 92. The memory 91 is adapted to store instructions or programs executable by the processor 90. Processor 90 may be a stand-alone microprocessor or a collection of one or more microprocessors. Thus, the processor 90 implements the processing of data and the control of other devices by executing instructions stored by the memory 91 to perform the method flows of embodiments of the present invention as described above. The bus 92 connects the above components together, as well as to a display controller 93 and a display device and input/output (I/O) device 94. Input/output (I/O) devices 94 may be a mouse, keyboard, modem, network interface, touch input device, motion sensing input device, printer, and other devices known in the art. Typically, the input/output devices 94 are connected to the system through input/output (I/O) controllers 95.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, apparatus (device) or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may employ a computer program product embodied on one or more computer-readable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations of methods, apparatus (devices) and computer program products according to embodiments of the application. It will be understood that each flow in the flow diagrams can be implemented by computer program instructions.
These computer program instructions may be stored in a computer-readable memory that can direct a computer or other programmable vehicle dispatch device to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows.
These computer program instructions may also be provided to a processor of a general purpose computer, special purpose computer, embedded processor or other programmable vehicle scheduling apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable vehicle scheduling apparatus, create means for implementing the functions specified in the flowchart flow or flows.
Another embodiment of the invention is directed to a non-transitory storage medium storing a computer-readable program for causing a computer to perform some or all of the above-described method embodiments.
That is, as can be understood by those skilled in the art, all or part of the steps in the method for implementing the embodiments described above may be implemented by a program instructing related hardware, where the program is stored in a storage medium and includes several instructions to enable a device (which may be a single chip, a chip, or the like) or a processor (processor) to execute all or part of the steps of the method described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.
The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention, and various modifications and changes may be made by those skilled in the art. Any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.
Claims (14)
1. A vehicle scheduling method, characterized in that the method comprises:
determining a target area, wherein the target area comprises a plurality of parking areas;
determining an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, wherein the attribute vector sequence comprises a plurality of attribute vectors corresponding to each time slice in the time slice sequence, the attribute vectors comprise attribute values corresponding to each parking area, and the attribute values are used for representing vehicle demand or vehicle supply of the corresponding parking areas;
inputting the attribute vector sequence into a plurality of graph convolution layers respectively so as to output a plurality of corresponding graph characteristics respectively;
determining an attribute value range corresponding to each parking area in a target time slice according to the time slice sequence and each graph characteristic, wherein the interval duration between the target time slice and the last time slice in the time slice sequence is one period of the time slice sequence;
and carrying out vehicle dispatching on each parking area according to the corresponding attribute value range.
2. The method of claim 1, wherein the attribute value is a vehicle demand;
the determining the attribute vector sequence corresponding to the target region according to the preset periodic time slice sequence includes:
respectively acquiring the number of vehicles in each parking area in the target area at the starting time point and the ending time point of each time slice in a preset periodic time slice sequence;
for each parking area, determining a corresponding vehicle demand according to the number of vehicles at the starting time point and the number of vehicles at the ending time point of each time slice;
and determining an attribute vector sequence according to the vehicle demand corresponding to each parking area in each time slice in the target area.
3. The method according to claim 1, characterized in that the attribute value is a vehicle supply amount;
the determining the attribute vector sequence corresponding to the target region according to the preset periodic time slice sequence includes:
acquiring the number of vehicles in each parking area in the target area at the starting time point of each time slice in a preset periodic time slice sequence;
for each parking area, determining a corresponding vehicle supply amount according to the number of vehicles at each time slice starting time point;
and determining an attribute vector sequence according to the vehicle supply quantity corresponding to each parking area in each time slice in the target area.
4. The method according to claim 1, wherein the inputting the sequence of attribute vectors into a plurality of graph convolution layers respectively to output a plurality of corresponding graph features respectively is specifically:
and inputting the attribute vector sequence into a first graph volume layer, a second graph volume layer and a third graph volume layer respectively so as to output corresponding first graph characteristics, second graph characteristics and third graph characteristics.
5. The method of claim 4, wherein the first graph convolutional layer is determined according to training of a set of Euclidean distance vectors corresponding to the target region, and each Euclidean distance vector in the set of Euclidean distance vectors is determined according to a distance between every two parking regions in the target region.
6. The method according to claim 4, wherein the second map convolutional layer is determined according to region information vector set training corresponding to the target region, and each region information vector in the region information vector set is determined according to the number of information points around each two parking regions in the target region.
7. The method according to claim 4, wherein the third graph convolution layer is determined according to training of a set of road network information vectors corresponding to the target region, and each road network information vector in the set of road network information vectors is determined according to road information around each two parking regions in the target region.
8. The method of claim 1, wherein determining the attribute value range corresponding to each parking area in the target time slice according to the time slice sequence and each map feature comprises:
inputting the time slice sequence and each graph feature into an attention mechanism layer to determine a corresponding attribute feature vector;
and after the attribute feature vectors are subjected to normalization processing, inputting the normalized attribute feature vectors into a full connection layer so as to determine the attribute value range corresponding to each parking area in a target time slice.
9. The method of claim 8, wherein the inputting the sequence of time slices and each of the graph features into an attention mechanism layer to determine a corresponding attribute feature vector comprises:
performing point multiplication on the time slice sequence and a preset position code, and then merging the time slice sequence and each graph feature to determine a corresponding attention input vector;
determining a corresponding attention feature vector according to a plurality of attention weight matrixes in the attention mechanism layer and an attention input vector;
an attribute feature vector is determined from each of the attention feature vectors and a logistic regression function.
10. The method of claim 8, wherein the normalizing the attribute feature vector and inputting the normalized attribute feature vector into a full link layer to determine the attribute value range corresponding to each parking area in a target time slice comprises:
inputting the attribute feature vector into a first normalization layer to determine a first feature vector;
inputting the first eigenvector into a feedforward layer to determine a second eigenvector;
inputting the first feature vector and the second feature vector into a second normalization layer to determine a third feature vector;
and inputting the third feature vector into a full-connection layer to determine the attribute value range corresponding to each parking area in the target time slice.
11. The method of claim 10, wherein inputting the third eigenvector into a fully-connected layer to determine the range of attribute values corresponding to each parking region in a target time slice comprises:
inputting the third feature vector into a full-connection layer to calculate and determine a plurality of predicted values corresponding to each parking area according to a quantile regression function and a plurality of preset quantiles;
and determining the attribute value range corresponding to the target time slice of each parking area according to a preset output rule and a plurality of corresponding predicted values.
12. A vehicle dispatching device, comprising:
the parking system comprises an area determining module, a parking area determining module and a parking area determining module, wherein the area determining module is used for determining a target area, and the target area comprises a plurality of parking areas;
the attribute determining module is used for determining an attribute vector sequence corresponding to the target area according to a preset periodic time slice sequence, wherein the attribute vector sequence comprises a plurality of attribute vectors corresponding to each time slice in the time slice sequence, the attribute vectors comprise attribute values corresponding to each parking area, and the attribute values are used for representing the vehicle demand or the vehicle supply quantity of the corresponding parking area;
the graph characteristic determining module is used for respectively inputting the attribute vector sequence into a plurality of graph convolution layers so as to respectively output a plurality of corresponding graph characteristics;
the prediction module is used for determining an attribute value range corresponding to each parking area in a target time slice according to the time slice sequence and each graph characteristic, and the interval duration between the target time slice and the last time slice in the time slice sequence is one period of the time slice sequence;
and the scheduling module is used for scheduling the vehicles in each parking area according to the corresponding attribute value range.
13. A computer readable storage medium storing computer program instructions, which when executed by a processor implement the method of any one of claims 1-11.
14. An electronic device comprising a memory and a processor, wherein the memory is configured to store one or more computer program instructions, wherein the one or more computer program instructions are executed by the processor to implement the method of any of claims 1-11.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010997098.8A CN112101804B (en) | 2020-09-21 | 2020-09-21 | Vehicle scheduling method and device, readable storage medium and electronic equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010997098.8A CN112101804B (en) | 2020-09-21 | 2020-09-21 | Vehicle scheduling method and device, readable storage medium and electronic equipment |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112101804A true CN112101804A (en) | 2020-12-18 |
CN112101804B CN112101804B (en) | 2021-11-02 |
Family
ID=73756394
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010997098.8A Active CN112101804B (en) | 2020-09-21 | 2020-09-21 | Vehicle scheduling method and device, readable storage medium and electronic equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112101804B (en) |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160210860A1 (en) * | 2013-06-26 | 2016-07-21 | Bayerische Motoren Werke Aktiengesellschaft | Method for processing measurement data of a vehicle in order to determine the start of a search for a parking space |
CN108776852A (en) * | 2018-06-22 | 2018-11-09 | 北京京东金融科技控股有限公司 | Without stake vehicle dispatching method and system |
CN111242292A (en) * | 2020-04-26 | 2020-06-05 | 北京全路通信信号研究设计院集团有限公司 | OD data prediction method and system based on deep space-time network |
CN111260424A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Information processing method and device |
CN111340543A (en) * | 2020-02-25 | 2020-06-26 | 上海大学 | Fine-grained taxi taking demand prediction method based on attention mechanism |
CN111476588A (en) * | 2019-01-24 | 2020-07-31 | 北京嘀嘀无限科技发展有限公司 | Order demand prediction method and device, electronic equipment and readable storage medium |
CN111612122A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Method and device for predicting real-time demand and electronic equipment |
-
2020
- 2020-09-21 CN CN202010997098.8A patent/CN112101804B/en active Active
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20160210860A1 (en) * | 2013-06-26 | 2016-07-21 | Bayerische Motoren Werke Aktiengesellschaft | Method for processing measurement data of a vehicle in order to determine the start of a search for a parking space |
CN108776852A (en) * | 2018-06-22 | 2018-11-09 | 北京京东金融科技控股有限公司 | Without stake vehicle dispatching method and system |
CN111260424A (en) * | 2018-11-30 | 2020-06-09 | 北京嘀嘀无限科技发展有限公司 | Information processing method and device |
CN111476588A (en) * | 2019-01-24 | 2020-07-31 | 北京嘀嘀无限科技发展有限公司 | Order demand prediction method and device, electronic equipment and readable storage medium |
CN111612122A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Method and device for predicting real-time demand and electronic equipment |
CN111340543A (en) * | 2020-02-25 | 2020-06-26 | 上海大学 | Fine-grained taxi taking demand prediction method based on attention mechanism |
CN111242292A (en) * | 2020-04-26 | 2020-06-05 | 北京全路通信信号研究设计院集团有限公司 | OD data prediction method and system based on deep space-time network |
Also Published As
Publication number | Publication date |
---|---|
CN112101804B (en) | 2021-11-02 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US11107101B2 (en) | Computing estimated value of providing service among geographical regions | |
CN109559147B (en) | Advertisement traffic prediction method, device, server and readable storage medium | |
CN108010378B (en) | Parking lot berth occupancy rate prediction method, device, equipment and storage medium | |
CN109983306A (en) | For generating the method and system of parking-related data | |
CN108230725A (en) | Method and apparatus are recommended in parking | |
US20210174270A1 (en) | Rideshare vehicle demand forecasting device, method for forecasting rideshare vehicle demand, and storage medium | |
CN109243619A (en) | Generation method, device and the computer readable storage medium of prediction model | |
CN113050856B (en) | Interaction method, interaction device, storage medium and electronic equipment | |
Antonio et al. | An advanced planner for urban freight delivering | |
CN112985442B (en) | Driving path matching method, readable storage medium and electronic device | |
WO2018146622A1 (en) | Dynamic selection of geo-based service options in a network system | |
CN111261017A (en) | Map fence display method and device | |
Pahlevani et al. | A cluster-based algorithm for home health care planning: A case study in Australia | |
CN114004077B (en) | Traffic simulation conversion method, device, computer equipment and storage medium | |
CN112991008B (en) | Position recommendation method and device and electronic equipment | |
JP2009015739A (en) | Traffic information distribution server, traffic information distribution system, and traffic information distribution method | |
CN114036411A (en) | Route planning method, device, equipment and medium | |
CN112101804B (en) | Vehicle scheduling method and device, readable storage medium and electronic equipment | |
CN112101671B (en) | Region attribute prediction method and device, readable storage medium and electronic equipment | |
Rahman et al. | TRIBUTE: Trip-based urban transportation emissions model for municipalities | |
JP2019071017A (en) | Road construction scheduling device and road construction scheduling program | |
Rahaman | Context-aware mobility analytics and trip planning | |
CN107680024A (en) | A kind of moving line creation method, device and computer-readable recording medium | |
WO2021230808A1 (en) | Server and method of determining an advanced booking fee for an advance booking | |
CN111626588A (en) | Vehicle delivery area determining method, device, server and storage medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |