CN117575677A - Training method, device, equipment and storage medium of mapping relation prediction model - Google Patents

Training method, device, equipment and storage medium of mapping relation prediction model Download PDF

Info

Publication number
CN117575677A
CN117575677A CN202311586030.0A CN202311586030A CN117575677A CN 117575677 A CN117575677 A CN 117575677A CN 202311586030 A CN202311586030 A CN 202311586030A CN 117575677 A CN117575677 A CN 117575677A
Authority
CN
China
Prior art keywords
data
charging
prediction model
mapping relation
vehicle
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.)
Pending
Application number
CN202311586030.0A
Other languages
Chinese (zh)
Inventor
王震坡
龙超华
刘鹏
祁春玉
夏智卿
石文童
贾自艳
韩伟
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Bitnei Corp ltd
Original Assignee
Beijing Bitnei Corp ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Beijing Bitnei Corp ltd filed Critical Beijing Bitnei Corp ltd
Priority to CN202311586030.0A priority Critical patent/CN117575677A/en
Publication of CN117575677A publication Critical patent/CN117575677A/en
Pending legal-status Critical Current

Links

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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
    • G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • 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
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • 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
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Strategic Management (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Development Economics (AREA)
  • Data Mining & Analysis (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • General Business, Economics & Management (AREA)
  • Evolutionary Computation (AREA)
  • Game Theory and Decision Science (AREA)
  • Artificial Intelligence (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Water Supply & Treatment (AREA)
  • General Health & Medical Sciences (AREA)
  • Human Resources & Organizations (AREA)
  • Primary Health Care (AREA)
  • Tourism & Hospitality (AREA)
  • Charge And Discharge Circuits For Batteries Or The Like (AREA)

Abstract

The invention relates to the technical field of data processing, and discloses a training method, a training device, training equipment and a training storage medium of a mapping relation prediction model, wherein the training method comprises the following steps: acquiring charging behavior data of a vehicle identification code and charging order data of a user serial number; constructing gap characteristics between the charging behavior data and the charging order data based on the charging behavior data and the charging order data; based on the gap characteristics, training the initial mapping relation prediction model to enable the mapping relation prediction model to learn the association relation between the first characteristic value and the second characteristic value of the key index, and obtaining the target mapping relation prediction model. According to the invention, the gap characteristics are constructed by the gap of the corresponding data dimension between the single-time charging behavior data of the vehicle and the single-time data of the charging pile, the association relation between the gap characteristics is learned by using the initial model, the model with the capability of predicting the mapping relation is obtained through training, the matching between the vehicle identification code and the user serial number is realized, and the subsequent data can be conveniently used.

Description

Training method, device, equipment and storage medium of mapping relation prediction model
Technical Field
The invention relates to the technical field of data processing, in particular to a training method, a training device, training equipment and a training storage medium of a mapping relation prediction model.
Background
Charging is a relatively high-frequency and key link in the process of a vehicle owner, and influences the decision of a user on purchasing a new energy vehicle. Meanwhile, the charging station operators have urgent hopes to improve the single pile utilization rate and further improve the station profitability.
In the related technology, the main body data forms of the current 'vehicle-pile-road-network' are different and mutually isolated, and the electric vehicle has strong flexibility in running, charging and other use behaviors, and has a complex rule, so that the current sensing and predicting capability on charge-discharge adjustable resources is poor, and the large-scale electric vehicle is difficult to support to participate in efficient guiding and scheduling of the electric vehicle to the network V2G.
Disclosure of Invention
In view of the above, the invention provides a training method, a training device and a training storage medium for a mapping relation prediction model, so as to solve the problem that the data between the existing charging operators and the platform are difficult to match, and the subsequent data is inconvenient to use.
In a first aspect, the present invention provides a training method of a mapping relation prediction model, where the method includes:
Acquiring charging behavior data of a vehicle identification code and charging order data of a user serial number;
constructing difference features between the charging behavior data and the charging order data based on the charging behavior data and the charging order data, wherein the difference features comprise feature combinations corresponding to a plurality of key indexes, the features comprise first feature values and second feature values, the first feature values are feature values of the key indexes in the charging behavior data, and the second feature values are feature values of the key indexes in the charging order data;
based on the gap characteristics, training the initial mapping relation prediction model to enable the mapping relation prediction model to learn the association relation between the first characteristic value and the second characteristic value of the key index, and obtaining a target mapping relation prediction model, wherein the target mapping relation prediction model is used for outputting a matching result of the vehicle identification code and the user serial number according to the association relation.
According to the invention, the gap characteristics are constructed by the gap of the corresponding data dimension between the single charging behavior data of the vehicle and the single data of the charging pile, the association relation between the gap characteristics is learned by using the initial model, the model with the capability of predicting the mapping relation is obtained through training, the matching between the vehicle identification code and the user serial number can be realized through the mapping relation prediction model, the follow-up 'vehicle-pile-road-network' efficient collaborative interaction electric vehicle network V2G service is facilitated, and the use experience of the user is improved.
In an alternative embodiment, based on the charging behavior data and the charging order data, a gap feature between the charging behavior data and the charging order data is constructed, comprising:
taking the vehicle identification code and the user serial number as a standard set;
correlating the charging behavior data with the charging order data to construct a training sample;
and calculating to obtain the difference characteristic between the charging behavior data and the charging order data based on the training sample.
In the method, the gap characteristics are obtained through calculation by correlating the charging behavior data of the vehicle end with the charging order data of the charging pile end, so that the vehicle end charging behavior data and the charging order data of the charging pile end are convenient to train by taking the subsequent charging order data as training samples.
In an alternative embodiment, the charging behavior data and the charging order data are associated to construct a training sample, including:
correlating the charging behavior data with the charging order data to obtain a charging data set with potential matching relation;
judging whether the current charging data in the charging data set contains a vehicle identification code or not;
when the current charging data in the charging data set contains a vehicle identification code, recording the current data as positive sample data;
When the current charging data in the charging data set does not contain the vehicle identification code, recording the current data as negative sample data;
and combining the positive sample data with the negative sample data to obtain a training sample.
In the mode, the charging behavior data and the order data are associated, positive and negative samples are divided, so that samples for model training are obtained, the charging behaviors with the matching relationship potentially can be selected conveniently, and the mapping relationship prediction model can be trained conveniently by using training samples later.
In an alternative embodiment, the gap feature comprises: the difference between the order charge start time and the vehicle charge start time, the difference between the order charge end time and the vehicle charge end time, the difference between the order charge longitude and the vehicle longitude, the difference between the order charge latitude and the vehicle latitude, the difference between the order charge electric quantity and the vehicle charge electric quantity, the difference between the residual electric quantity of the vehicle at the order charge start time and the residual electric quantity at the vehicle charge start time, and the difference between the residual electric quantity of the vehicle at the order charge end time and the residual electric quantity at the vehicle charge end time are stored in the charge behavior data.
In the mode, through the gap characteristics, various aspects of the charging behavior are more comprehensively represented, so that the data of the vehicle end and the charging pile end can be comprehensively matched, and the matching accuracy is further improved.
In an alternative embodiment, training the initial mapping relation prediction model based on the gap feature to obtain a target mapping relation prediction model includes:
dividing the gap features into different gap features of the belonging areas based on longitude and latitude;
and training the initial mapping relation prediction model by utilizing the difference characteristics of the same areas to obtain a target mapping relation prediction model.
In the mode, by utilizing the gap characteristics belonging to the same region, the training by using data with too low matching probability is avoided, and the accuracy of model prediction and the efficiency of model training can be improved.
In an alternative embodiment, the method further comprises:
obtaining the matching probability between the vehicle identification code and the user serial number which are output by the target mapping relation prediction model for a plurality of times;
and optimizing the target mapping relation prediction model based on the vehicle identification code, the user serial number and the matching probability between the vehicle identification code and the user serial number to obtain an optimized target mapping relation prediction model.
In the mode, the historical matching data are counted, and the model is further optimized based on a long-term matching result, so that the accuracy and recall rate of the model for mapping relation prediction are improved.
In an alternative embodiment, optimizing the target mapping relation prediction model based on the vehicle identification code, the user serial number, and the probability of matching between the vehicle identification code and the user serial number includes:
calculating to obtain a matching gap characteristic based on the vehicle identification code, the user serial number, the matching probability between the vehicle identification code and the user serial number and the matching times;
and optimizing the mapping relation prediction model based on the matching gap characteristics.
In the mode, the matching result and the matching times are used as input through counting the matching times of the model in a period of time, so that the model is further optimized, and the accuracy of model prediction is further improved.
In a second aspect, the present invention provides a training apparatus for a mapping relation prediction model, where the apparatus includes:
the data acquisition module is used for acquiring charging behavior data of the vehicle identification code and charging order data of the user serial number;
the feature construction module is used for constructing difference features between the charging behavior data and the charging order data based on the charging behavior data and the charging order data, wherein the difference features comprise feature combinations corresponding to a plurality of key indexes, the features comprise first feature values and second feature values, the first feature values are feature values of the key indexes in the charging behavior data, and the second feature values are feature values of the key indexes in the charging order data;
The model training module is used for training the initial mapping relation prediction model based on the difference characteristic so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model, wherein the target mapping relation prediction model is used for outputting a matching result of the vehicle identification code and the user serial number according to the association relation.
In a third aspect, the present invention provides a computer device comprising: the memory and the processor are in communication connection with each other, the memory stores computer instructions, and the processor executes the computer instructions, so as to execute the training method of the mapping relation prediction model of the first aspect or any implementation manner corresponding to the first aspect.
In a fourth aspect, the present invention provides a computer-readable storage medium having stored thereon computer instructions for causing a computer to execute the training method of the mapping relation prediction model of the first aspect or any one of the embodiments corresponding thereto.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a training method of a mapping relation prediction model according to an embodiment of the present invention.
Fig. 2 is a logic diagram of a single-pass mapping relationship matching algorithm according to an embodiment of the present invention.
FIG. 3 is a flow chart of another method for training a mapping prediction model according to an embodiment of the present invention.
Fig. 4 is a flowchart of a training method of a mapping relation prediction model according to an embodiment of the present invention.
Fig. 5 is a block diagram of a training apparatus of a mapping relation prediction model according to an embodiment of the present invention.
Fig. 6 is a schematic diagram of a hardware structure of a computer device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments of the present invention. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
In the related technology, the main body data forms of the current 'vehicle-pile-road-network' are different and mutually isolated, and the electric vehicle has strong flexibility in running, charging and other use behaviors, and has a complex rule, so that the current sensing and predicting capability on charge-discharge adjustable resources is poor, and the large-scale electric vehicle is difficult to support to participate in efficient guiding and scheduling of the electric vehicle to the network V2G.
In order to solve the above-mentioned problems, in the embodiments of the present invention, a training method for a mapping relation prediction model is provided for a computer device, and it should be noted that an execution body of the training method may be a training device for the mapping relation prediction model, and the training device may be implemented by software, hardware or a combination of software and hardware to form part or all of the computer device, where the computer device may be a terminal, a client, or a server, and the server may be a server, or may be a server cluster formed by multiple servers. In the following method embodiments, the execution subject is a computer device.
The computer equipment in the embodiment is suitable for establishing a use scene of a shared account system for each link of the new energy automobile in an industrial chain. According to the training method of the mapping relation prediction model, the difference feature is built through the difference of the corresponding data dimension between the single-time charging behavior data of the vehicle and the single-time data of the charging pile, the association relation between the difference features is learned by the initial model, the model with the capability of predicting the mapping relation is obtained through training, the matching between the vehicle identification code and the user serial number can be realized through the mapping relation prediction model, the follow-up 'vehicle-pile-road-network' efficient collaborative interaction electric vehicle network V2G service is facilitated, and the user experience is improved.
According to an embodiment of the present invention, there is provided a training method embodiment of a mapping relation prediction model, it should be noted that the steps shown in the flowchart of the drawings may be performed in a computer system such as a set of computer executable instructions, and that although a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in an order different from that shown or described herein.
In this embodiment, a training method of a mapping relation prediction model is provided, which may be used in the above-mentioned computer device, and fig. 1 is a flowchart of a training method of a mapping relation prediction model according to an embodiment of the present invention, as shown in fig. 1, where the flowchart includes the following steps:
step S101, acquiring charging behavior data of a vehicle identification code and charging order data of a user serial number.
In an example, the charging order data is provided by a charging operator, consisting essentially of: (1) user serial number Id of charging user, (2) charging start vehicle battery state of charge SOC, (3) charging end vehicle battery state of charge SOC, (4) charging start time, (5) charging end time, (6) charging station longitude, (7) charging station latitude, and (8) charge amount.
The charging behavior data is vehicle behavior data collected by a vehicle end and comprises data such as running, charging and the like. In the charging behavior data, the data dimension provided by the charging operator is covered. Mainly comprises the following steps: (1) a vehicle identification code, (2) a vehicle charge start vehicle battery state of charge SOC, (3) a vehicle charge end vehicle battery state of charge SOC, (4) a vehicle charge start time, (5) a vehicle charge end time, (6) a vehicle charge station longitude, (7) a vehicle charge station latitude, and (8) an amount of charge of the vehicle.
Step S102, based on the charging behavior data and the charging order data, a gap feature between the charging behavior data and the charging order data is constructed.
In the embodiment of the invention, the gap features comprise feature combinations corresponding to a plurality of key indexes, the features comprise first feature values and second feature values, the first feature values are feature values of the key indexes in charging behavior data, and the second feature values are feature values of the key indexes in charging order data.
In one example, the gap feature is based on a common data dimension, the gap feature being made, the main gap feature comprising: 1) start_time_diff: charging start time in order-vehicle charging start time, 2) end_time_diff: charging end time in order-vehicle charging end time, 3) lng_diff: order charge longitude-vehicle longitude, 4) lat_diff: order charging latitude-vehicle latitude, 5) energy_diff: order charge amount-vehicle charge amount, 6) start_soc_diff: residual amount of vehicle at order charge start time-residual amount of vehicle charge start time i, 7) end_soc_diff: residual amount of vehicle soc at vehicle charge end time-residual amount of vehicle charge end time soc.
Step S103, training the initial mapping relation prediction model based on the difference features so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model.
In one example, the number of samples is relatively large because fewer features are available in the charging behavior and some of the features have some null values. The tree model LightGBM can be well adapted to the scene, has strong interpretability, and can obtain good accuracy, so that the Lightgbm model is selected for training. The parameters are tuned by grid search (GridSearch), the best key parameters are as follows:
since model prediction involves a large number of Cartesian product cross-product calculations, the amount of online prediction data is large, and thus the calculation efficiency of the model is an important problem. The main idea of performance optimization is to reduce the number of matching calculations through data binning to improve efficiency. The main idea of the data barrel is to divide the data into different grids according to longitude and latitude, and only the data in one grid is subjected to matching calculation, so that the efficiency is improved.
In an implementation scenario, fig. 2 is a logic schematic diagram of a single-shot mapping relation matching algorithm according to an embodiment of the present invention, and as shown in fig. 2, a process of a single-shot mapping relation matching model includes: acquiring order data in one day, selecting one piece of order data, judging whether a user serial number UserID of a user in the order data is in a mapping table, wherein the mapping table is used for representing a matching relation between a vehicle identification code and the user serial number, and if the order data is in the mapping table and the confidence coefficient is more than 90%, indicating that the vehicle identification code and the user serial number in the order data are in a one-to-one correspondence relation, performing next order data acquisition. If the order data is not in the mapping table or the order data is in the mapping table but the confidence coefficient is less than or equal to 90%, the order data is combined with the charging behavior data of the vehicle in one day, whether the order time and the vehicle starting time error in the order data and the charging behavior data are within 300 seconds is judged, and when the order time and the vehicle starting time error in the order data and the charging behavior data are within 300 seconds, feature calculation of the order data and the charging behavior data is carried out, a feature calculation result is input into a mapping relation prediction model for prediction, a matching result of a vehicle identification code and a user serial number obtained by model prediction is obtained, and the matching result of the vehicle identification code and the user serial number is updated into the mapping table. And performing effect evaluation on the matching result of the vehicle identification code and the user serial number in one day, and updating the mapping table when the accuracy is higher than x%. And when the accuracy is not higher than x%, optimizing the model, and re-running the data set to update the matching result of the model.
According to the training method of the mapping relation prediction model, the difference feature is built through the difference of the corresponding data dimension between the single-time charging behavior data of the vehicle and the single-time data of the charging pile, the association relation between the difference features is learned by the initial model, the model with the capability of predicting the mapping relation is obtained through training, matching between the vehicle identification code and the user serial number can be achieved through the mapping relation prediction model, the follow-up 'vehicle-pile-road-network' efficient collaborative interaction electric vehicle network V2G service is facilitated, and the user experience is improved.
In this embodiment, a training method of a mapping relation prediction model is provided, which may be used in the above-mentioned computer device, and fig. 3 is a flowchart of another training method of a mapping relation prediction model according to an embodiment of the present invention, as shown in fig. 3, where the flowchart includes the following steps:
in step S301, charging behavior data of the vehicle identification code and charging order data of the user serial number are obtained. Please refer to step S101 in the embodiment shown in fig. 1 in detail, which is not described herein.
Step S302, based on the charging behavior data and the charging order data, a gap feature between the charging behavior data and the charging order data is constructed.
Specifically, the step S302 includes:
in step S3021, the vehicle identification code and the user serial number are set as criteria.
In step S3022, the charging behavior data and the charging order data are correlated to construct a training sample.
In some alternative embodiments, step S3022 includes:
and a step a1, associating the charging behavior data with the charging order data to obtain a charging data set with potential matching relation.
And a2, judging whether the current charging data in the charging data set contains a vehicle identification code.
And a step a3, when the current charging data in the charging data set contains the vehicle identification code, recording the current data as positive sample data.
And a4, when the current charging data in the charging data set does not contain the vehicle identification code, recording the current data as negative sample data.
And a5, combining the positive sample data with the negative sample data to obtain a training sample.
In an example, the construction of the training samples may include: 1) Time range: and selecting a charging order with more charging behaviors and rich samples in a time period as a data sample. 2) The < UserID, vin > set is collected as a standard set. 3) By a certain rule (e.g.: order and vehicle in the same city, order start time and vehicle charge start time error within 300 seconds), and all charging behaviors potentially having matching relationships are selected.
Based on the data of all the charging behaviors which are potentially in matching relation, a matching label of a sample is constructed by combining a standard set (a standard set data is provided by a charging operator, and the data comprises corresponding results of userID and vin codes), and if the vehicle vin and the data appear in the standard set, the matching label is marked as 1 and is used as a positive sample, otherwise, the matching label is marked as 0 and is used as a negative sample.
In the mode, the charging behavior data and the order data are associated, positive and negative samples are divided, so that samples for model training are obtained, the charging behaviors with the matching relationship potentially can be selected conveniently, and the mapping relationship prediction model can be trained conveniently by using training samples later.
In step S3023, based on the training samples, a gap characteristic between the charging behavior data and the charging order data is calculated.
In some alternative embodiments, the gap feature comprises: the difference between the order charge start time and the vehicle charge start time, the difference between the order charge end time and the vehicle charge end time, the difference between the order charge longitude and the vehicle longitude, the difference between the order charge latitude and the vehicle latitude, the difference between the order charge electric quantity and the vehicle charge electric quantity, the difference between the residual electric quantity of the vehicle at the order charge start time and the residual electric quantity at the vehicle charge start time, and the difference between the residual electric quantity of the vehicle at the order charge end time and the residual electric quantity at the vehicle charge end time are stored in the charge behavior data.
In one example, the main gap features include: 1) start_time_diff: charging start time in order-vehicle charging start time, 2) end_time_diff: charging end time in order-vehicle charging end time, 3) lng_diff: order charge longitude-vehicle longitude, 4) lat_diff: order charging latitude-vehicle latitude, 5) energy_diff: order charge amount-vehicle charge amount, 6) start_soc_diff: residual amount of vehicle at order charge start time-residual amount of vehicle charge start time i, 7) end_soc_diff: residual amount of vehicle soc at vehicle charge end time-residual amount of vehicle charge end time soc.
In the mode, through the gap characteristics, various aspects of the charging behavior are more comprehensively represented, so that the data of the vehicle end and the charging pile end can be comprehensively matched, and the matching accuracy is further improved.
Step S303, training the initial mapping relation prediction model based on the gap characteristics so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model.
Specifically, the step S303 includes:
In step S3031, the gap features are divided into different gap features of the belonging areas based on the longitude and latitude.
In an example, the constructed samples are subjected to vehicle longitude and latitude coordinate conversion, converted into the same coordinate system, kept consistent with the longitude and latitude of the order, and prepared for the feature engineering. Map coordinate systems are of many kinds, and the following types are common:
(1) High longitude and latitude coordinate system (84 coordinate system)
The longitude and latitude high coordinate system (84 coordinate system) is a coordinate system composed of longitude (latitude), latitude (altitude), and altitude (alttude), also called LLA coordinate system. It can be said that it is the most widely used coordinate system.
(2) GCJ-02: the coordinate system of the national surveying and mapping geographic information bureau is also called as a Mars coordinate system. National regulations, internet maps must be encrypted for the first time at home using at least GCJ-02. The coordinate system is the most widely used coordinate system in China, such as hundred degrees, goldd and Tencentrated maps.
(3) Other special coordinate systems: is generally calculated from Mars coordinates by an offset algorithm.
The values of the different poi points in different coordinate systems are different, and each charging operator can adopt different coordinate systems to store the position information. The function of the coordinate conversion of longitude and latitude is to establish the connection between various different maps and realize the information sharing and exchange between the different maps. The transformation of the coordinate system can be realized by calling a function, and the transformation method of the coordinate system is not limited in the invention.
Step S3032, training the initial mapping relation prediction model by utilizing the difference characteristics of the same area to obtain a target mapping relation prediction model.
In one example, the number of samples is relatively large because fewer features are available in the charging behavior and some of the features have some null values. The tree model LightGBM can be well adapted to the scene, has strong interpretability, and can obtain good accuracy, so that the Lightgbm model is selected for training. The parameters are tuned by grid search (GridSearch), the best key parameters are as follows:
since model prediction involves a large number of Cartesian product cross-product calculations, the amount of online prediction data is large, and thus the calculation efficiency of the model is an important problem. The main idea of performance optimization is to reduce the number of matching calculations through data binning to improve efficiency. The main idea of the data barrel is to divide the data into different grids according to longitude and latitude, and only the data in one grid is subjected to matching calculation, so that the efficiency is improved.
In the mode, by utilizing the gap characteristics belonging to the same region, the training by using data with too low matching probability is avoided, and the accuracy of model prediction and the efficiency of model training can be improved.
According to the training method of the mapping relation prediction model, the gap characteristics are obtained through calculation by correlating the charging behavior data of the vehicle end with the charging order data of the charging pile end, so that the vehicle end charging behavior data and the charging order data of the charging pile end are used as training samples for training. The charging behavior data and the order data are associated, positive and negative samples are divided, so that samples for model training are obtained, all potential charging behaviors with matching relations are conveniently selected, and the mapping relation prediction model is conveniently trained by using training samples. Through the gap characteristics, various aspects of the charging behavior are more comprehensively represented, the data of the vehicle end and the charging pile end are conveniently and comprehensively matched, and the matching accuracy is further improved.
In this embodiment, a training method of a mapping relation prediction model is provided, which may be used in the above-mentioned computer device, and fig. 4 is a flowchart of a training method of a mapping relation prediction model according to another embodiment of the present invention, as shown in fig. 4, where the flowchart includes the following steps:
in step S401, charging behavior data of the vehicle identification code and charging order data of the user serial number are obtained. Please refer to step S301 in the embodiment shown in fig. 3 in detail, which is not described herein.
Step S402, based on the charging behavior data and the charging order data, a gap feature between the charging behavior data and the charging order data is constructed. Please refer to step S302 in the embodiment shown in fig. 3 in detail, which is not described herein.
Step S403, training the initial mapping relation prediction model based on the gap characteristics so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model. Please refer to step S303 in the embodiment shown in fig. 3 in detail, which is not described herein.
Step S404, obtaining the matching probability between the vehicle identification code and the user serial number output by the target mapping relation prediction model for a plurality of times.
Step S405, optimizing the target mapping relation prediction model based on the vehicle identification code, the user serial number and the matching probability between the vehicle identification code and the user serial number, and obtaining the optimized target mapping relation prediction model.
Specifically, the step S405 includes:
in step S4051, a matching gap feature is calculated based on the vehicle identification code, the user serial number, the matching probability between the vehicle identification code and the user serial number, and the number of matches.
And step S4052, optimizing the mapping relation prediction model based on the matching gap characteristics.
In an example, a process of optimizing and building a model using historical charging behavior may include: 1. feature structure
Model matching based on single behavior data is carried out every day, and a single mapping table < userID, VIN, matching probability > of every day is produced. The effect of single matching needs to be improved, so that the matching times of each < userID, VIN > in a period of time (for example, in a year) are counted, and the more the matching times are, the greater the accuracy of the matching relation is. Thus, the following data statistics need to be performed:
1) The number of times a user matches is noted: user_counts; 2) The number of times the vehicle was matched is noted as: vin_counts; 3) The number of times the user matches the vehicle is noted as: user_vin_counts; 4) The user order number is recorded as: user_order_counts; 5) Confidence of match: the ratio of the number of matches (user_vin_counts) between the user and the vehicle to the number of matches (user_counts) for the user ID is expressed as: conf_rate; 6) Matching support degree: the ratio of the number of matches between the user and the vehicle (user_vin_counts) to the number of matches between the user and the vehicle (vin_counts) is expressed as: a supp_rate; 7) Matching coverage: the ratio of the number of times the user matches the vehicle (user_vin_counts) to the number of orders of the user (user_order_counts) is recorded as: math_covRate.
For statistics of the ratio class, a large number law is considered in calculation, and a conf_rate (confidence rate) is calculated as an example, and a specific calculation formula is as follows:
a＝user_vin_counts+1
b＝1+user_counts-user_vin_counts
wherein: user_counts are all matches for user ID, user_vin_counts are the matches for user and vehicle.
2. Sample construction: the sample at the time of single matching (each order charge and vehicle charge matching) is selected as a sample modeled based on historical data, and the statistics characteristics (user_counts, vin_counts, user_vin_counts, user_order_counts, conf_rate, supply_rate and Math_covrate) calculated in the first step are added to the original sample data to supplement the sample data.
3. Model training: after sample data were obtained, training was performed using the following supplements: (1) first the sample was taken as 7:3, dividing a training set and a testing set; (2) Obtaining optimal super parameters on the training set through grid search (GridSearh); (3) after the super parameters are determined, generating a model instance; (4) Model training is carried out based on the training set, and the effect of the model on the training set is obtained; (5) Evaluating the test set by the trained model to obtain an evaluation result of the model on the test set; (6) And judging whether the fitting is performed or not based on the effects on the training set and the test set.
And updating the final mapping table by day, and updating the mapping table based on the model matching result generated every day.
1) If the user_id and vin in the model matching result are not in the mapping table, the whole piece of information is directly inserted. user_counts, user_vin_counts, user_order_counts are all set to 1.
2) If the user_id in the model matching result is in the mapping table and the vin is not in the mapping table, updating all information related to the user_id, directly inserting the information related to the vin, and setting the vin_counts to be 1.
3) If user_id in the model matching result is not in the mapping table but the vin is in the mapping table, updating all information related to the vin, and setting user_counts to 1.
4) If the user_id and the vin in the model matching result are both in the mapping table, updating all information related to the user_id and the vin.
5) And recalculating the matching confidence, the matching support and the matching coverage rate every day, and recalculating the model to generate a new mapping table.
The sample construction process also comprises the identification of abnormal values of the sample, the sample source is a vehicle vin submitted by a user through the product function, the data cannot be ensured to be completely correct in the mode, the condition that the user upgrades and changes the vehicle or rents the vehicle can also occur, and the quality of the sample cannot be ensured. Because the quality of the sample has a great influence on the effect of the model, strategies are taken to perform sample purification. The method specifically comprises the following steps:
1. Positive outlier removal:
1) And carrying out data distribution statistics on the positive example data in the sample to obtain the quantile value of each feature.
2) Defining abnormal values according to the box diagram, and removing characteristics larger than outlier up Or less than outlier down Is a sample of (a).
outlier up ＝(quantile 75％ -quantile 25％ )*3+quantile 75％
outlier down ＝quantile 25％ -(quantile 75％ -quantile 25％ )*3
Therein, outlier up Threshold of maximum value, outlier down Is the minimum threshold, quaternie 75％ To take the value of the upper quartile, quaternion 25％ The lower quartile is given value.
Since the outlier data is the data deviating from the normal sample data distribution, the outlier sample data can be predicted by a trained model, and should not be normal samples under normal conditions, if the predicted result is that the proportion of the normal samples is small, it is feasible to remove the outlier.
2. Negative example removes outliers: 1) And carrying out data distribution statistics on the positive example data in the sample to obtain a data distribution value of each feature, for example: mean, quartile, etc.
2) The true label value is 0 (negative sample), theoretically the number of matches between UserID and VIN (user_vin_counts) will be small, so if the user_vin_counts are large, the confidence (conf_rate), support (supp_rate) and coverage (math_covrate) will all be high (a value greater than 75% quantiles in the positive case), and these data will be removed. After the abnormal value is removed, the model is retrained, and the model effect is obviously improved. The influence of the data quality on the model effect is large, abnormal values are removed based on a data statistics method, and the model effect is further improved.
In the mode, the matching result and the matching times are used as input through counting the matching times of the model in a period of time, so that the model is further optimized, and the accuracy of model prediction is further improved.
According to the training method of the mapping relation prediction model, statistics is carried out on historical matching data, and the model is further optimized based on long-term matching results, so that the accuracy and recall rate of the model on mapping relation prediction are improved. And the matching result and the matching times are used as input through counting the matching times in a period of time, so that the model is further optimized, and the accuracy of model prediction is further improved.
The embodiment also provides a training device for the mapping relation prediction model, which is used for implementing the above embodiment and the preferred implementation manner, and is not described in detail. As used below, the term "module" may be a combination of software and/or hardware that implements a predetermined function. While the means described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and contemplated.
The present embodiment provides a training device for a mapping relation prediction model, as shown in fig. 5, including:
the data acquisition module 501 is configured to acquire charging behavior data of a vehicle identification code and charging order data of a user serial number. Please refer to step S101 in the embodiment shown in fig. 1 in detail, which is not described herein.
The feature construction module 502 is configured to construct a gap feature between the charging behavior data and the charging order data based on the charging behavior data and the charging order data, where the gap feature includes feature combinations corresponding to a plurality of key indicators, and the feature includes a first feature value and a second feature value, the first feature value is a feature value of the key indicator in the charging behavior data, and the second feature value is a feature value of the key indicator in the charging order data. Please refer to step S102 in the embodiment shown in fig. 1 in detail, which is not described herein.
The model training module 503 is configured to train the initial mapping relation prediction model based on the gap feature, so that the mapping relation prediction model learns an association relation between the first feature value and the second feature value of the key index, and obtains a target mapping relation prediction model, where the target mapping relation prediction model is configured to output a matching result of the vehicle identification code and the user serial number according to the association relation. Please refer to step S103 in the embodiment shown in fig. 1 in detail, which is not described herein.
In some alternative implementations, the feature construction module 502 includes:
and the standard set construction unit is used for taking the vehicle identification code and the user serial number as standard sets.
And the training sample construction unit is used for correlating the charging behavior data with the charging order data to construct a training sample.
And the gap characteristic calculation unit is used for calculating and obtaining the gap characteristic between the charging behavior data and the charging order data based on the training sample.
In some alternative embodiments, the training sample construction unit comprises:
and the data association subunit is used for associating the charging behavior data with the charging order data to obtain a charging data set with potential matching relation.
And the vehicle identification code judging subunit is used for judging whether the current charging data in the charging data set contains the vehicle identification code.
And the positive sample construction subunit is used for recording the current data as positive sample data when the current charging data in the charging data set contains the vehicle identification code.
And the negative sample construction subunit is used for recording the current data as negative sample data when the current charging data in the charging data set does not contain the vehicle identification code.
And the training sample construction subunit is used for combining the positive sample data with the negative sample data to obtain a training sample.
In some alternative embodiments, the gap feature comprises: the difference between the order charge start time and the vehicle charge start time, the difference between the order charge end time and the vehicle charge end time, the difference between the order charge longitude and the vehicle longitude, the difference between the order charge latitude and the vehicle latitude, the difference between the order charge electric quantity and the vehicle charge electric quantity, the difference between the residual electric quantity of the vehicle at the order charge start time and the residual electric quantity at the vehicle charge start time, and the difference between the residual electric quantity of the vehicle at the order charge end time and the residual electric quantity at the vehicle charge end time are stored in the charge behavior data.
In some alternative embodiments, model training module 503 includes:
the region dividing unit is used for dividing the gap features into different gap features of the belonging regions based on longitude and latitude.
And the model training unit is used for training the initial mapping relation prediction model by utilizing the difference characteristics of the same region to obtain a target mapping relation prediction model.
In some optional embodiments, the training apparatus of the mapping relation prediction model further includes:
And the matching data acquisition unit is used for acquiring the matching probability between the vehicle identification code and the user serial number which are output by the target mapping relation prediction model for a plurality of times.
The model optimizing unit is used for optimizing the target mapping relation prediction model based on the vehicle identification code, the user serial number and the matching probability between the vehicle identification code and the user serial number, and obtaining an optimized target mapping relation prediction model.
In some alternative embodiments, the model optimization unit includes:
and the matching gap feature construction subunit is used for calculating and obtaining the matching gap feature based on the vehicle identification code, the user serial number, the matching probability between the vehicle identification code and the user serial number and the matching times.
And the model optimization subunit is used for optimizing the mapping relation prediction model based on the matching gap characteristics.
Further functional descriptions of the above respective modules and units are the same as those of the above corresponding embodiments, and are not repeated here.
The training device of the mapping relation prediction model in this embodiment is presented in the form of functional units, where the units refer to ASIC (Application Specific Integrated Circuit ) circuits, processors and memories executing one or more software or fixed programs, and/or other devices that can provide the above functions.
The embodiment of the invention also provides a computer device, which is provided with the training device of the mapping relation prediction model shown in the figure 5.
Referring to fig. 6, fig. 6 is a schematic structural diagram of a computer device according to an alternative embodiment of the present invention, as shown in fig. 6, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting the various components, including high-speed interfaces and low-speed interfaces. The various components are communicatively coupled to each other using different buses and may be mounted on a common motherboard or in other manners as desired. The processor may process instructions executing within the computer device, including instructions stored in or on memory to display graphical information of the GUI on an external input/output device, such as a display device coupled to the interface. In some alternative embodiments, multiple processors and/or multiple buses may be used, if desired, along with multiple memories and multiple memories. Also, multiple computer devices may be connected, each providing a portion of the necessary operations (e.g., as a server array, a set of blade servers, or a multiprocessor system). One processor 10 is illustrated in fig. 6.
The processor 10 may be a central processor, a network processor, or a combination thereof. The processor 10 may further include a hardware chip, among others. The hardware chip may be an application specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general-purpose array logic, or any combination thereof.
Wherein the memory 20 stores instructions executable by the at least one processor 10 to cause the at least one processor 10 to perform the methods shown in implementing the above embodiments.
The memory 20 may include a storage program area that may store an operating system, at least one application program required for functions, and a storage data area; the storage data area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include high-speed random access memory, and may also include non-transitory memory, such as at least one magnetic disk storage device, flash memory device, or other non-transitory solid-state storage device. In some alternative embodiments, memory 20 may optionally include memory located remotely from processor 10, which may be connected to the computer device via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
Memory 20 may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as flash memory, hard disk, or solid state disk; the memory 20 may also comprise a combination of the above types of memories.
The computer device also includes a communication interface 30 for the computer device to communicate with other devices or communication networks.
The embodiments of the present invention also provide a computer readable storage medium, and the method according to the embodiments of the present invention described above may be implemented in hardware, firmware, or as a computer code which may be recorded on a storage medium, or as original stored in a remote storage medium or a non-transitory machine readable storage medium downloaded through a network and to be stored in a local storage medium, so that the method described herein may be stored on such software process on a storage medium using a general purpose computer, a special purpose processor, or programmable or special purpose hardware. The storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, a solid state disk or the like; further, the storage medium may also comprise a combination of memories of the kind described above. It will be appreciated that a computer, processor, microprocessor controller or programmable hardware includes a storage element that can store or receive software or computer code that, when accessed and executed by the computer, processor or hardware, implements the methods illustrated by the above embodiments.
Although embodiments of the present invention have been described in connection with the accompanying drawings, various modifications and variations may be made by those skilled in the art without departing from the spirit and scope of the invention, and such modifications and variations fall within the scope of the invention as defined by the appended claims.

Claims (10)

1. A method for training a predictive model of a mapping relationship, the method comprising:
acquiring charging behavior data of a vehicle identification code and charging order data of a user serial number;
constructing a gap feature between the charging behavior data and the charging order data based on the charging behavior data and the charging order data, wherein the gap feature comprises a feature combination corresponding to a plurality of key indexes, the feature comprises a first feature value and a second feature value, the first feature value is a feature value of the key index in the charging behavior data, and the second feature value is a feature value of the key index in the charging order data;
training an initial mapping relation prediction model based on the gap characteristics so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model, wherein the target mapping relation prediction model is used for outputting a matching result of a vehicle identification code and the user serial number according to the association relation.
2. The method of claim 1, wherein the constructing a gap feature between the charging behavior data and the charging order data based on the charging behavior data and the charging order data comprises:
taking the vehicle identification code and the user serial number as a standard set;
correlating the charging behavior data with the charging order data to construct a training sample;
and calculating the gap characteristic between the charging behavior data and the charging order data based on the training sample.
3. The method of claim 2, wherein the correlating the charging behavior data and the charging order data to construct training samples comprises:
correlating the charging behavior data with the charging order data to obtain a charging data set with potential matching relation;
judging whether the current charging data in the charging data set contains the vehicle identification code or not;
when the current charging data in the charging data set contains the vehicle identification code, recording the current data as positive sample data;
when the current charging data in the charging data set does not contain the vehicle identification code, recording the current data as negative sample data;
And combining the positive sample data with the negative sample data to obtain a training sample.
4. The method of claim 2, wherein the gap feature comprises: the difference between the order charge start time and the vehicle charge start time, the difference between the order charge end time and the vehicle charge end time, the difference between the order charge longitude and the vehicle longitude, the difference between the order charge latitude and the vehicle latitude, the difference between the order charge electric quantity and the vehicle charge electric quantity, the difference between the residual electric quantity of the vehicle at the order charge start time and the residual electric quantity at the vehicle charge start time, and the difference between the residual electric quantity of the vehicle at the order charge end time and the residual electric quantity at the vehicle charge end time are stored in the charge behavior data.
5. The method of claim 4, wherein training the initial mapping prediction model based on the gap feature to obtain the target mapping prediction model comprises:
dividing the gap features into different gap features of the belonging areas based on longitude and latitude;
and training the initial mapping relation prediction model by utilizing the difference characteristics of the same areas to obtain a target mapping relation prediction model.
6. The method according to claim 1, wherein the method further comprises:
obtaining the matching probability between the vehicle identification code and the user serial number which are output by the target mapping relation prediction model for a plurality of times;
and optimizing the target mapping relation prediction model based on the vehicle identification code, the user serial number and the matching probability between the vehicle identification code and the user serial number to obtain an optimized target mapping relation prediction model.
7. The method of claim 6, wherein optimizing the target mapping prediction model based on the vehicle identification code, the user serial number, a probability of a match between the vehicle identification code and the user serial number, comprises:
calculating to obtain a matching gap characteristic based on the vehicle identification code, the user serial number, the matching probability between the vehicle identification code and the user serial number and the matching times;
and optimizing the mapping relation prediction model based on the matching gap characteristics.
8. A training device for a mapping relation prediction model, the device comprising:
the data acquisition module is used for acquiring charging behavior data of the vehicle identification code and charging order data of the user serial number;
The characteristic construction module is used for constructing a gap characteristic between the charging behavior data and the charging order data based on the charging behavior data and the charging order data, wherein the gap characteristic comprises a characteristic combination corresponding to a plurality of key indexes, the characteristic comprises a first characteristic value and a second characteristic value, the first characteristic value is a characteristic value of the key index in the charging behavior data, and the second characteristic value is a characteristic value of the key index in the charging order data;
the model training module is used for training the initial mapping relation prediction model based on the gap characteristics so that the mapping relation prediction model learns the association relation between the first characteristic value and the second characteristic value of the key index to obtain a target mapping relation prediction model, wherein the target mapping relation prediction model is used for outputting a matching result of the vehicle identification code and the user serial number according to the association relation.
9. A computer device, comprising:
a memory and a processor, the memory and the processor being communicatively connected to each other, the memory having stored therein computer instructions, the processor executing the computer instructions to perform the training method of the mapping relation prediction model of any one of claims 1 to 7.
10. A computer-readable storage medium having stored thereon computer instructions for causing a computer to perform the training method of the mapping relation prediction model according to any one of claims 1 to 7.
CN202311586030.0A 2023-11-24 2023-11-24 Training method, device, equipment and storage medium of mapping relation prediction model Pending CN117575677A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202311586030.0A CN117575677A (en) 2023-11-24 2023-11-24 Training method, device, equipment and storage medium of mapping relation prediction model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202311586030.0A CN117575677A (en) 2023-11-24 2023-11-24 Training method, device, equipment and storage medium of mapping relation prediction model

Publications (1)

Publication Number Publication Date
CN117575677A true CN117575677A (en) 2024-02-20

Family

ID=89889695

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202311586030.0A Pending CN117575677A (en) 2023-11-24 2023-11-24 Training method, device, equipment and storage medium of mapping relation prediction model

Country Status (1)

Country Link
CN (1) CN117575677A (en)

Similar Documents

Publication Publication Date Title
CN106557829B (en) Method and device for acquiring region with unmatched demand and transport capacity in vehicle service
CN109342989B (en) Error analysis method and device for electric energy meter
CN111950225B (en) Chip layout method and device, storage medium and electronic equipment
EP3349126B1 (en) Method, device, storage medium, and apparatus for automatically discovering fuel station poi
WO2021208079A1 (en) Method and apparatus for obtaining power battery life data, computer device, and medium
US20160131501A1 (en) Correlating meter data to detect mismappings
CN110634021A (en) Big data based vehicle estimation method, system, device and readable storage medium
CN112613627A (en) Vehicle maintenance information pushing method, device, system, equipment and storage medium
CN114879070A (en) Battery state evaluation method and related equipment
CN114693045A (en) Power conversion station address selection method and system, electronic equipment and storage medium
CN117575677A (en) Training method, device, equipment and storage medium of mapping relation prediction model
CN112782584B (en) Method, system, medium and device for predicting remaining usage amount of battery electric quantity
CN115617633A (en) Icing monitoring terminal maintenance method and related equipment
CN114896482A (en) Model training and energy supplementing intention recognition method, device, equipment and medium
CN116050554A (en) Scenic spot passenger flow volume prediction method and device, computing equipment and storage medium
CN110163465B (en) Pipeline inspection scheduling method and device
CN108990072B (en) Multi-operator site planning method and device
CN112800102A (en) Alarm correlation calculation method and device and calculation equipment
CN112895967B (en) Method, system, medium, and device for predicting remaining service time of battery replacement mileage
WO2022156743A1 (en) Feature construction method and apparatus, model training method and apparatus, and device and medium
CN116451882B (en) Method for predicting carbon emission and related equipment
CN115714708A (en) Method and device for determining non-broadband user and readable storage medium
CN112785465B (en) Knowledge point pushing method, system, medium and device
CN113722506A (en) Intelligent knowledge point identification method, system, intelligent equipment and storage medium
CN115056677A (en) Method and device for testing charging map data and electronic equipment

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