CN111626527A - Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle - Google Patents

Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle Download PDF

Info

Publication number
CN111626527A
CN111626527A CN202010522503.0A CN202010522503A CN111626527A CN 111626527 A CN111626527 A CN 111626527A CN 202010522503 A CN202010522503 A CN 202010522503A CN 111626527 A CN111626527 A CN 111626527A
Authority
CN
China
Prior art keywords
charging
electric automobile
day
power
time
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
Application number
CN202010522503.0A
Other languages
Chinese (zh)
Other versions
CN111626527B (en
Inventor
秦文萍
史文龙
姚宏民
景祥
朱云杰
高蒙楠
韩肖清
贾燕冰
任春光
王磊
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.)
Taiyuan University of Technology
Original Assignee
Taiyuan University of Technology
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 Taiyuan University of Technology filed Critical Taiyuan University of Technology
Priority to CN202010522503.0A priority Critical patent/CN111626527B/en
Publication of CN111626527A publication Critical patent/CN111626527A/en
Application granted granted Critical
Publication of CN111626527B publication Critical patent/CN111626527B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

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
    • G06Q10/067—Enterprise or organisation modelling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • 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
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E40/00—Technologies for an efficient electrical power generation, transmission or distribution
    • Y02E40/70—Smart grids as climate change mitigation technology in the energy generation sector
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
    • Y02T90/00—Enabling technologies or technologies with a potential or indirect contribution to GHG emissions mitigation
    • Y02T90/10—Technologies relating to charging of electric vehicles
    • Y02T90/16—Information or communication technologies improving the operation of electric vehicles
    • Y02T90/167—Systems integrating technologies related to power network operation and communication or information technologies for supporting the interoperability of electric or hybrid vehicles, i.e. smartgrids as interface for battery charging of electric vehicles [EV] or hybrid vehicles [HEV]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S10/00—Systems supporting electrical power generation, transmission or distribution
    • Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
    • Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
    • Y04S30/00—Systems supporting specific end-user applications in the sector of transportation
    • Y04S30/10—Systems supporting the interoperability of electric or hybrid vehicles
    • Y04S30/12—Remote or cooperative charging

Abstract

The invention discloses a smart power grid deep learning scheduling method considering a schedulable electric vehicle fast/slow charging/discharging form, which belongs to the field of regional smart power grid operation, and performs power supply power optimization distribution by taking total operation cost as a target function at a day-ahead scheduling stage; simulating load fluctuation and a day-ahead scheduling plan in the in-day pre-scheduling stage to serve as deep learning network input samples, inputting simulation generated prediction data into a regional intelligent power grid model in the in-day pre-scheduling stage, and taking controllable unit scheduling data in a model training stage scheduling plan as output samples of a deep learning network. Training a regional intelligent power grid in-day scheduling model based on a deep learning network through input samples and output samples; obtaining a predicted value of the next scheduling time of the load through ultra-short term prediction; and inputting the predicted value and the day-ahead scheduling plan into a regional intelligent power grid day scheduling model together to obtain a day scheduling value of the controllable unit. The invention solves the problems that the distributed power supply of the regional intelligent power grid, the electric automobile and the load prediction have errors, the daily economic dispatching of the regional intelligent power grid is difficult to realize, and the like.

Description

Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle
Technical Field
The invention relates to the field of regional intelligent power grid dispatching, in particular to an intelligent power grid deep learning dispatching method considering a schedulable electric vehicle fast/slow charging/discharging mode.
Background
Power scheduling is an important task in modern energy management systems, and under the constraints of power generation, power transmission and operation, the system operation economy becomes an important target of power scheduling. In the traditional power dispatching, only thermal power generating units are involved, and a dispatching center makes a starting plan of the units by taking the lowest coal consumption as a target according to the unit combination state and system parameters of the last dispatching day. With the global energy crisis and the environmental issues being raised, the vigorous development of renewable energy has become a common consensus among countries in the world. Modern renewable energy sources such as wind energy, water energy, solar energy, tidal energy, geothermal energy and the like are rapidly developed in the last decade, and wind power generation and photovoltaic power generation become main driving forces for the high-speed development of the whole renewable energy industry. With the development of battery technology, plug-in electric vehicles (electric vehicles) have been developed rapidly in recent years, and have various effects on conventional power dispatching. The rapidly-increased charging demand of the electric automobile brings new pressure to the power generation capacity of a power grid, the power generation capacity expansion cost is improved, and the power generation economic benefit is reduced. The development of a vehicle-to-grid (V2G) technology for an electric vehicle can enable the electric vehicle to better participate in a power grid dispatching process. After a large number of electric vehicles are connected to a power grid, if the power grid can discharge in the peak load period of the power grid through dispatching of the electric vehicles and charge in the valley load period of the power grid, the peak clipping and valley filling effects can be achieved on the power grid, and the consumption capacity of a power system on renewable energy is improved. Therefore, in the power dispatching problem, loads need to be distributed not only among traditional thermal power generating units but also among electric vehicles and renewable energy sources, and the regional smart grid dispatching method becomes a hot spot of current research. In the existing research, the adverse effects caused by restraining wind power output and load fluctuation and uncertainty by adjusting the charging and discharging power of the electric vehicle and the accuracy of the scheduling strategy depending on the day-ahead scheduling are not considered, so that the scheduling plan of the power grid is easily affected.
Disclosure of Invention
The invention provides an intelligent power grid deep learning scheduling method considering the fast/slow charging/discharging form of a schedulable electric vehicle, and solves the problems that adverse effects caused by wind power output, load fluctuation and uncertainty are restrained by adjusting the charging/discharging power of the electric vehicle, a scheduling strategy depends on the accuracy of day-ahead scheduling, and the scheduling plan of a power grid is easily influenced.
The invention is realized by the following technical scheme: the intelligent power grid dispatching system mainly comprises a power supply end and a load end, as shown in fig. 2, wherein the power generation end comprises a V2G system, a thermal power generating unit and a wind farm, and the load end comprises a conventional load, a large number of disordered electric vehicles and a large number of schedulable electric vehicle charging loads. In order to reduce the starting and stopping cost of the thermal power generating unit, in the day-ahead scheduling process, the largest two conventional thermal power generating units in the thermal power generating unit are always in an opening state, so that the thermal power generating unit is divided into a thermal power generating unit I and a thermal power generating unit II, the thermal power generating unit I is composed of the two largest conventional thermal power generating units, and the thermal power generating unit II is composed of other units. A smart power grid deep learning scheduling method considering a schedulable electric vehicle fast/slow charging/discharging mode comprises a day-ahead scheduling stage, a model training stage and a day-in scheduling stage;
the day-ahead scheduling stage:
(1) segmenting according to hours, dividing 1 day into 24 time periods, and taking the power output and absorption of each distributed unit in each time period as fixed values;
(2) calling the predicted wind power plant generating power, the charging power of the disordered charging electric vehicle and the conventional load fluctuation condition data in each time period in the future day;
(3) establishing a mathematical model of the electric automobile:
a. classifying the electric automobile:
according to the charging characteristics of the electric automobile, the electric automobile is divided into three categories: the first type is an unordered charging electric automobile, such as an electric taxi, an electric bus, an electric logistics car, a large number of private cars which do not participate in power grid dispatching and the like, and the public service type vehicle has the important characteristics that the average driving time per day is relatively long, the requirements on the charging speed and the charging time are high, and the public service type vehicle cannot participate in power grid interaction; the second type is a fast-charging schedulable electric automobile, the third type is a slow-charging schedulable electric automobile, the two types of schedulable electric automobiles are mostly private cars, and the fast-charging schedulable electric automobile is characterized in that the fast-charging schedulable electric automobile is idle in most of the day, enough time can participate in power grid interaction, the fast-charging schedulable electric automobile can be used as a V2G power supply in the peak period of power utilization, and meanwhile, the charge-discharge characteristics of the fast-charging schedulable electric automobile are utilized to stabilize wind power fluctuation and reduce the running and start-stop costs of a unit. The influence of the charge and discharge process of the slow-charging schedulable electric automobile on the battery loss is small, but the scheduling response speed is low; the fast-charging schedulable electric automobile is more flexible in scheduling, but certain extra battery loss can be brought by the fast-charging characteristic.
b. Electric vehicle state matrix:
the state of any electric automobile at the end of driving is represented by a one-dimensional matrix:
Ω＝[L N SnSeTsTe]
wherein: l represents the type of the electric automobile load, and numbers 1, 2, 3, 4 and 5 respectively represent the disordered charging load, the slow-charging schedulable discharging load, the fast-charging schedulable charging load and the fast-charging schedulable discharging load of the electric automobile; n represents a charging and discharging identifier of the electric automobile, the charging mode is 1, the discharging mode is-1, and the rest moments are 0; snAnd SeRespectively representing the state of charge of the electric automobile when the electric automobile is stopped and the expected state of charge of a user when the electric automobile is off the network; t issAnd TeRespectively representing the network access time of the electric automobile and the network leaving time of a user;
c. unordered electric automobile load model:
the method comprises the following steps of fitting survey data of domestic vehicles in the United states in 2009 to obtain electric vehicle driving end time distribution and battery state of charge distribution before charging, and extracting the electric vehicle driving end time of the last time, the state of charge when the electric vehicle finishes driving and the daily driving mileage of an electric vehicle user in an electric vehicle state matrix by adopting a Monte Carol method to generate a disordered electric vehicle model:
the daily mileage of the electric vehicle user approximately follows the lognormal distribution, and the probability density function is as follows:
Figure BDA0002532655320000031
in the formula: d is the mileage, mu, of the automobiled＝3.019,σd＝1.123；
The SOC of the battery of the electric automobile and the driving mileage d thereof approximately satisfy the following linear relation:
E＝(1-d/D)×100％；
the last trip end time approximately follows the Weibull distribution as follows:
Figure BDA0002532655320000032
in the formula: k is a radical oftIs a shape parameter, ctIs a scale parameter;
setting the electric automobile to be charged only once a day, wherein the charging time starts from the end of the last driving; and the residual electric quantity of the electric automobile user is charged when 20% -50%, and the expected off-grid time of the user is uniformly set to be Te7, expected charge S when user is away from homeeUniform distribution of 80% -100% is obeyed;
d. the electric automobile schedulable charge-discharge load model:
at present, most electric private cars adopt lithium batteries, and considering that the service life of the batteries is attenuated and lost due to the fact that the electric cars are scheduled to be charged and discharged, the charging and discharging switching times of the electric cars are reduced as much as possible; in the set model, only one-time discharging scheduling is carried out on each automobile one day, charging efficiency and battery service life are considered for reducing the influence on the power distribution network, and schedulable charging and discharging capacity calculation is carried out on second and third types of electric automobiles:
as shown in fig. 1, it is a schematic diagram of an SOC time node of an electric vehicle; setting t0To t1The time period of (1) is the time node length of the SOC of the electric vehicle, wherein t and t are also includedlimThe two nodes, t0The moment when the electric automobile arrives at the charging place and is connected into the power grid is shown, and the electric quantity of the battery of the electric automobile is Q0(ii) a t represents the current time of the electric automobile, and the electric quantity of the battery of the electric automobile is Q at the momentt(ii) a From time t0The state of charge to t is known, and the period of time represents that the electric vehicle is scheduled by the power grid for a period of time; t is t1At the moment when the electric vehicle leaves the power grid, the state of charge of the electric vehicle is Q1(ii) a Then from t to t1The time interval is the time interval of the electric vehicle to be scheduled by the power grid; during the period, the electric automobile is subjected to discharge scheduling, and the discharge process is transferred from t to continue to tlimTime of day tlimThe method is characterized in that the discharge limit time point of the electric automobile is represented, the time point represents the maximum time when the electric automobile discharges under the condition of meeting the expected state of charge when the electric automobile user is off the network, and if the maximum time exceeds the discharge limit time pointWhen the electric automobile leaves at the time point, the battery power cannot meet the user requirement; electric automobile at tlimState of charge at time Qlim(ii) a From tlimTo t1The period of time is a period of time when the electric automobile needs to be charged after being discharged, and because the electric quantity of the battery of the electric automobile is in a lower level through the previous discharging, the electric automobile needs to be charged in the later period of time to meet the electric quantity required by the user when the user is off the network;
electric automobile discharge electrode time limit tlimThe calculation is as follows:
Figure BDA0002532655320000041
in the formula: t is the current time of the electric automobile; the battery charge quantity of the electric automobile at the time t is Q1；t1The moment when the electric automobile leaves the power grid; pcCharging power for the electric vehicle; pdDischarging power for the electric vehicle; csThe battery capacity of the electric automobile;
the schedulable discharge capacity of the electric automobile is calculated as follows:
Figure BDA0002532655320000042
the electric vehicle charging capacity is calculated as follows:
Pev,c＝Cs[Qt+Q1-Q0-Qlim]
extracting a schedulable electric automobile matrix from the electric automobile state matrix by using a Monte Carlo method, determining the number of schedulable electric automobiles, and determining the discharge limit time point t of each schedulable electric automobilelimThe schedulable charging and discharging capacity is added to the schedulable charging and discharging capacity curve of each schedulable electric automobile, and finally the schedulable electric automobile charging and discharging limit capacity is obtained;
(4) the optimization model established in the day-ahead scheduling stage comprises a target function with the lowest total operation cost of the regional intelligent power grid and a distributed power supply constraint condition;
a. the lowest objective function of the total operation cost of the regional intelligent power grid in the day-ahead scheduling stage takes the fuel cost, the starting and stopping cost of a conventional unit and the electricity purchasing and selling cost of the electric automobile into account, and the expression is as follows:
Figure BDA0002532655320000051
f is the total operation cost of the system in the day-ahead economic optimization scheduling stage; n is a radical of1The number of the thermal power generating units I,
Figure BDA0002532655320000052
a fuel cost function of the unit i in a time period t;
Figure BDA0002532655320000053
generating power of the unit i in a time period t;
Figure BDA0002532655320000054
starting and stopping a unit i at a time t; n is a radical of2The number of the thermal power generating units II is,
Figure BDA0002532655320000055
as a function of the fuel cost for unit i over time t,
Figure BDA0002532655320000056
the generated power of the unit i in the time period t,
Figure BDA0002532655320000057
is the starting and stopping state of the unit i in the time period t,
Figure BDA0002532655320000058
the method comprises the steps of obtaining a start-stop cost function of a unit i in a time period t;
Figure BDA0002532655320000059
scheduling Capacity for charging of a slow-charging schedulable electric vehicle at time t, μcCharging electric automobile in t periodThe price of electricity is set according to the price of electricity,
Figure BDA00025326553200000510
scheduling Capacity for discharging of a trickle charge schedulable electric vehicle at time t, μdThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure BDA00025326553200000511
scheduling the capacity of the electric vehicle for fast charging in the t time period;
Figure BDA00025326553200000512
scheduling capacity for discharging of fast-charging schedulable electric vehicle in t period ηbScheduling battery loss penalty cost for the fast-charging schedulable electric vehicle;
Figure BDA00025326553200000513
scheduling capacity for charging of the disordered electric vehicle at a time period t;
b. in order to ensure the safe and reliable operation of the regional intelligent power grid, each unit in the regional intelligent power grid needs to satisfy a certain equality constraint or inequality constraint condition in each time interval, and the method comprises the following steps:
i thermal power generating unit output upper and lower limit constraint and climbing constraint:
Figure BDA0002532655320000061
Figure BDA0002532655320000062
wherein the content of the first and second substances,
Figure BDA0002532655320000063
the minimum output and the maximum output of the unit i are respectively set;
Figure BDA0002532655320000064
respectively limiting the power of climbing up and down slopes of the unit i;
II thermal power generating units output upper and lower limit constraints, start and stop constraints and climbing constraints:
Figure BDA0002532655320000065
Figure BDA0002532655320000066
Figure BDA0002532655320000067
Figure BDA0002532655320000068
wherein the content of the first and second substances,
Figure BDA0002532655320000069
the minimum output and the maximum output of the unit i are respectively set;
Figure BDA00025326553200000610
respectively the continuous on-off time and the minimum on-off time of the unit i;
Figure BDA00025326553200000611
respectively limiting the power of climbing up and down slopes of the unit i;
scheduling capacity constraint of the slow charging electric automobile:
Figure BDA00025326553200000612
Figure BDA00025326553200000613
Figure BDA00025326553200000614
wherein the content of the first and second substances,
Figure BDA00025326553200000615
are respectively slowly chargedThe upper and lower limits of the charging capacity of the electric automobile;
Figure BDA00025326553200000616
the upper limit of the discharge capacity of the slow charging electric automobile;
fourthly, the schedulable capacity of the fast-charging electric automobile is restrained:
Figure BDA00025326553200000617
Figure BDA00025326553200000618
Figure BDA00025326553200000619
wherein the content of the first and second substances,
Figure BDA00025326553200000620
respectively setting the upper limit and the lower limit of the charging capacity of the fast charging electric automobile;
Figure BDA00025326553200000621
the upper limit of the discharge capacity of the quick-charging electric automobile is defined;
system load balancing constraint:
Figure BDA0002532655320000071
wherein the content of the first and second substances,
Figure BDA0002532655320000072
the wind power plant output power is in a time period t;
Figure BDA0002532655320000073
normal load power for a period t;
(5) according to the objective function with the lowest total operation cost of the regional intelligent power grid in the step (4) and the constraint condition of the distributed power supply; solved by the yalcip software module of MATLAB: the method comprises the following steps that in each period of time in the future, the operation state of a thermal power generating unit I, the operation state of a thermal power generating unit II, the charge and discharge power of a slow-charge schedulable electric automobile and the charge and discharge power of a fast-charge schedulable electric automobile are set; scheduling the total running cost of the system day ahead;
II, model training stage:
(1) in the model training phase scheduling plan, 15 minutes are taken as unit time intervals, and the whole day is divided into 96 time intervals;
(2) simulating and generating electric field generating power, disordered charging electric vehicle charging power and conventional load fluctuation condition data at each time period in a day by a disturbance increasing method;
(3) combining the electric field generating power, the disordered charging electric vehicle charging power and the conventional load fluctuation condition of each time period in the day generated by simulation with the day-ahead scheduling data to be used as a deep learning network input sample;
(4) in the model training stage, the thermal power generating unit II and the slow-charging schedulable electric automobile execute a day-ahead scheduling plan;
(5) the operation and maintenance cost of the thermal power generating unit I, the operation and maintenance cost of the fast-charging schedulable electric automobile and the operation and maintenance cost of the disordered charging electric automobile are considered by optimizing the objective function in the model training stage, and the expression is as follows:
Figure BDA0002532655320000074
wherein, FXThe total cost of system operation in the model training stage; n is a radical of1The number of the thermal power generating units I,
Figure BDA0002532655320000075
as a function of the fuel cost for unit i over time t,
Figure BDA0002532655320000076
the generated power of the unit i in the time period t,
Figure BDA0002532655320000077
starting and stopping a unit i at a time t;
Figure BDA0002532655320000078
for the charge scheduling capacity of the fast charge schedulable electric automobile in the time period t,
Figure BDA0002532655320000079
capacity of discharge scheduling for fast-charging schedulable electric vehicle at time t, ηb-XScheduling Battery loss penalty cost, μ for fast Charge schedulable electric vehiclescCharging price of electric vehicle in t period, mudThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure BDA00025326553200000710
scheduling capacity for charging of the disordered electric vehicle at a time period t;
(6) in order to ensure safe and reliable operation of the regional intelligent power grid, in a model training phase, all units in the regional intelligent power grid meet the constraint conditions and are the same as those in a day-ahead scheduling phase;
(7) and (3) solving a model training stage through a yalcip software module of MATLAB according to the optimization objective function in the step (5) and the constraint condition in the step (6): the thermal power generating unit I simulates generating power, and the fast charging adjustable electric automobile simulates charging and discharging power; the system operation total cost and the like in the model training stage are obtained, and the solved simulated power generation power of the thermal power generating unit I and the simulated charge and discharge power of the fast-charging schedulable electric automobile are used as output samples of the deep learning network;
a. in the model training and scheduling stage, the input samples in the deep learning network training samples are as follows:
Figure BDA0002532655320000081
in the formula: n is a radical ofinputSimulating input samples for the deep learning network within a time period t; t is a time scale;
Figure BDA0002532655320000082
simulating the difference between the wind power plant power and the forecast power of the wind power plant in the day before in the time period t;
Figure BDA0002532655320000083
simulating the difference between the conventional load and the predicted conventional load in the day ahead for the time period t;
Figure BDA0002532655320000084
the difference between the load of the simulated disordered charging electric vehicle in the time period t and the predicted load of the disordered charging electric vehicle in the day ahead is calculated;
Figure BDA0002532655320000085
simulating wind farm power for a time period t;
Figure BDA0002532655320000086
simulating a conventional load for a time period t;
Figure BDA0002532655320000087
simulating the load of the disordered charging electric vehicle in the time period t;
Figure BDA0002532655320000088
scheduling capacity for discharging of the slow-charging schedulable PEV at a time t;
Figure BDA0002532655320000089
discharging scheduling capacity of the slow charging schedulable PEV in a t period;
Figure BDA00025326553200000810
discharging scheduling capacity of the PEV in the t period is scheduled for the quick charging of the day-ahead scheduling stage;
Figure BDA00025326553200000811
the charging scheduling capacity of the PEV in the t time period can be scheduled for the quick charging in the day-ahead scheduling stage;
Figure BDA00025326553200000812
generating power of a unit i in the thermal power generating unit I in a day-ahead scheduling stage in a t period;
Figure BDA00025326553200000813
generating power of a unit i in a thermal power unit II in a time period t;
Figure BDA00025326553200000814
predicting power for a wind farm day ahead within a time period t;
Figure BDA00025326553200000815
predicting a conventional load for a day ahead within a time period t;
Figure BDA00025326553200000816
the load of the disorderly charged electric automobile is predicted in the day ahead;
b. taking the scheduling data of the controllable units in the model training stage as output samples of deep learning network training, wherein the output samples in the deep learning network training samples are as follows:
Figure BDA00025326553200000817
in the formula: n is a radical ofoutputSimulating output samples for the deep learning network within the time period t;
Figure BDA00025326553200000818
simulating discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA0002532655320000091
the simulation charging power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA0002532655320000092
the method comprises the following steps that (1) the analog output power of a thermal power generating unit I in a time period t is obtained;
(8) repeating the steps (2) to (7), adding input samples and output samples, and training a deep learning network to obtain an intraday scheduling model;
thirdly, scheduling stage in a day:
(1) in the scheduling stage in the day, taking 15 minutes as a unit time interval, dividing the whole day into 96 time intervals;
(2) forecasting the generated power of the wind power plant, the charging power of the disordered electric vehicle and the conventional load fluctuation condition at each time interval in a day in an ultra-short period;
(3) inputting the next-time ultra-short term prediction data and a day-ahead scheduling plan into a day scheduling model to obtain the power generation power of the thermal power generating unit I and the charge-discharge power of the fast-charging schedulable electric automobile as next-time scheduling values;
the input data of the deep learning network are as follows:
Figure BDA0002532655320000093
in the formula: n is a radical ofS-inputInputting data in real time for the deep learning network within a time period t;
Figure BDA0002532655320000094
the difference between the ultra-short-term predicted power of the wind power plant in the day and the predicted power of the wind power plant in the day before is the time period t;
Figure BDA0002532655320000095
the difference between the ultra-short-term predicted conventional load in the day and the predicted conventional load before the day within the time interval t is obtained;
Figure BDA0002532655320000096
the difference between the super short-term predicted load of the unordered charging electric automobile in the day and the predicted load of the unordered charging electric automobile in the day in the time period t;
Figure BDA0002532655320000097
ultra-short-term power prediction is carried out on the wind power plant in the day within the time period t;
Figure BDA0002532655320000098
predicting the conventional load for the ultra-short term in the day within the time period t;
Figure BDA0002532655320000099
and predicting the load of the disorderly charged electric automobile in the day within the time period t in the ultra-short term.
The output data of the deep learning network is as follows:
Figure BDA00025326553200000910
in the formula: n is a radical ofS-outputReal-time output data of the deep learning network in a representation time period t;
Figure BDA00025326553200000911
representing the real-time discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA00025326553200000912
representing the real-time charging power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA00025326553200000913
representing the real-time output power of the thermal power generating unit I in a time period t;
(4) and in the day dispatching stage, the thermal power generating unit II and the slow charging dispatchable electric automobile execute a day-ahead dispatching plan.
Compared with the prior art, the invention has the following beneficial effects: the invention provides a smart power grid deep learning scheduling method considering a schedulable electric vehicle fast/slow charging/discharging mode, which comprises the following steps: (1) in a day-ahead scheduling stage, peak, valley and average electricity prices in all time periods are considered, and according to the output of a day-ahead wind power plant, the load of an unordered charging electric vehicle and the conventional load prediction, the total operation cost of a thermal power generating unit, a schedulable electric vehicle and the like is taken as a target function to carry out distributed power supply power optimization distribution in a regional intelligent power grid, so that the operation state of the intelligent power grid is more comprehensively processed;
(2) the method utilizes the regulation of the charging and discharging power of the electric automobile to inhibit the adverse effects caused by wind power output, load fluctuation and uncertainty, and divides the schedulable electric automobile into a fast-charging schedulable electric automobile and a slow-charging schedulable electric automobile. And determining the dispatching plans of the slow-charging schedulable electric automobile and the thermal power generating unit II in a day-ahead dispatching stage in consideration of the dispatching response speed of the slow-charging schedulable electric automobile and the reduction of the starting and stopping cost of the thermal power generating unit as far as possible. On the basis of a day-ahead scheduling plan, time scale in a day is further accurate, the characteristic of flexible scheduling of the fast-charging schedulable electric vehicle is fully exerted, a day scheduling model is used for scheduling the controllable units of the regional intelligent power grid, and stable operation of the regional intelligent power grid is guaranteed.
(3) According to the invention, by comparing the scheduling strategy based on the BP neural network, the economic scheduling strategy based on the LSTM deep learning network is closer to the scheduling result in the future, the scheduling cost is almost consistent, and the economy and the effectiveness of the scheduling strategy based on the LSTM deep learning network in the scheduling process of the smart grid are verified.
Drawings
FIG. 1 is a schematic diagram of an SOC time node of an electric vehicle according to the present invention.
Fig. 2 is a structure of a smart grid dispatching system according to the present invention.
FIG. 3 is a predicted wind farm output power curve from the day ahead according to the present invention.
Fig. 4 is a daily actual curve of the output power of the wind farm according to the invention.
Fig. 5 is a load curve of a day ahead predictive chaotic electric vehicle according to the present invention.
Fig. 6 is a daily actual load curve of the disordered electric vehicle according to the present invention.
FIG. 7 is a predicted daily conventional load demand curve in accordance with the present invention.
Fig. 8 is a daily actual curve of a normal load according to the present invention.
Fig. 9 is a thermal power generating unit ii day-ahead scheduling operation curve according to the present invention.
FIG. 10 is a slow charge dispatchable electric vehicle dispatch operating curve in accordance with the invention.
Fig. 11 is a thermal power generating unit i day-ahead scheduling operation curve according to the present invention.
Fig. 12 is a day-ahead scheduled operation curve of the fast-charging schedulable electric vehicle according to the present invention.
Fig. 13 is a daily scheduling operating curve of the fast-charging electric vehicle according to the present invention.
Fig. 14 is a scheduling operation curve of the thermal power generating unit in the day i according to the invention.
FIG. 15 is a flowchart of a smart grid deep learning scheduling strategy taking into account the schedulable electric vehicle fast/slow charge/discharge pattern in accordance with the present invention.
Fig. 16 is a flow chart of an algorithm for calculating a random charging electric vehicle according to the present invention.
Fig. 17 is a flowchart of a schedulable charge-discharge electric vehicle prediction according to the present invention.
Detailed Description
The present invention is further illustrated by the following specific examples.
The intelligent power grid dispatching system mainly comprises a power supply end and a load end, as shown in fig. 2, wherein the power generation end comprises a V2G system, a thermal power generating unit and a wind power plant, and the load end comprises a conventional load, a large number of disordered electric vehicles and a large number of schedulable electric vehicle charging loads. In order to reduce the starting and stopping cost of the thermal power generating unit, in the day-ahead scheduling process, the largest two conventional thermal power generating units in the thermal power generating unit are always in an opening state, so that the thermal power generating unit is divided into a thermal power generating unit I and a thermal power generating unit II, the thermal power generating unit I is composed of the two largest conventional thermal power generating units, and the thermal power generating unit II is composed of other units. According to the invention, a 10-machine system is taken as an example for analysis and calculation, the capacity of a thermal power machine assembling machine is 1668MW, wherein thermal power units 1 and 2 form a thermal power unit I, the installed capacity is 900MW, thermal power units 3-10 form a thermal power unit II, and the installed capacity is 768 MW. 16 wind power stations of 100MW are added in the original system, and the total installed capacity is 1600 MW. The number of regional electric vehicles is 30 thousands, namely 10 thousands of disorderly charged electric vehicles, 10 thousands of fast-charging schedulable electric vehicles and 10 thousands of slow-charging schedulable electric vehicles. In order to reduce the loss of the schedulable electric vehicle to the electric vehicle battery by participating in the scheduling, the following assumptions are made: the electric automobile is scheduled only once a day, and the charging time starts from the end of the last driving. The electric vehicle can be charged when the residual electric quantity of a user is 20% -50% (the user expects to uniformly set the off-grid time as T)e7) expected charge S when the user is away from homeeUniform distribution of (80% -100%) was followed. The parameters of the fast-charging schedulable electric automobile and the slow-charging schedulable electric automobile are shown in the table 1; the charge and discharge electricity prices of the electric vehicle are shown in table 2:
TABLE 1 electric vehicle parameter settings
Figure BDA0002532655320000111
TABLE 2 Charge and discharge electricity price of electric vehicle
Figure BDA0002532655320000121
A smart grid deep learning scheduling method considering a schedulable electric vehicle fast/slow charging/discharging mode comprises a day-ahead scheduling stage, a model training stage and a day-in scheduling stage as shown in FIG. 15;
the day-ahead scheduling stage:
(1) segmenting according to hours, dividing 1 day into 24 time periods, and taking the power output and absorption of each distributed unit in each time period as fixed values;
(2) calling the predicted wind power plant generating power, the charging power of the disordered charging electric vehicle and the conventional load fluctuation condition data in each time period in the future day; fig. 3 is a curve of output power of a wind power plant predicted in the day ahead, fig. 5 is a curve of load of a chaotic electric vehicle predicted in the day ahead, fig. 7 is a curve of demand of a conventional load predicted in the day ahead, fig. 10 is a curve of scheduling operation of a thermal power generating unit II in the day ahead, and fig. 11 is a curve of scheduling operation of a slow-charging schedulable electric vehicle;
(3) establishing a mathematical model of the electric automobile:
a. classifying the electric automobile:
according to the charging characteristics of the electric automobile, the electric automobile is divided into three categories: the first type is an unordered charging electric automobile, the second type is a fast-charging schedulable electric automobile, and the third type is a slow-charging schedulable electric automobile;
b. electric vehicle state matrix:
the state of any electric automobile at the end of driving is represented by a one-dimensional matrix:
Ω＝[L N SnSeTsTe]
wherein: l represents the electric vehicle load type and is represented by the number 1,2. 3, 4 and 5 respectively represent an electric automobile disordered charging load, a slow charging schedulable discharging load, a fast charging schedulable charging load and a fast charging schedulable discharging load; n represents a charging and discharging identifier of the electric automobile, the charging mode is 1, the discharging mode is-1, and the rest moments are 0; snAnd SeRespectively representing the state of charge of the electric automobile when the electric automobile is stopped and the expected state of charge of a user when the electric automobile is off the network; t issAnd TeRespectively representing the network access time of the electric automobile and the network leaving time of a user;
c. unordered electric automobile load model:
since electric utility vehicle loads are almost negligible compared to electric private cars, the present document focuses on the simulation of chaotic electric vehicle charging loads for private cars that do not participate in grid dispatching. Because the electric automobile is not popularized and applied in a large scale, the daily driving mileage and the last trip ending time of the domestic vehicle in the whole America are approximately subjected to lognormal distribution and Weibull distribution by analyzing the US fuel private car and driving data in 2009. Extracting the last driving end time of the electric automobile, the charge state when the electric automobile finishes driving and the daily driving mileage of an electric automobile user in the state matrix of the electric automobile by adopting a Monte Carol method to generate a disordered electric automobile model:
the daily mileage of the electric vehicle user approximately follows the lognormal distribution, and the probability density function is as follows:
Figure BDA0002532655320000131
in the formula: d is the mileage, mu, of the automobiled＝3.019,σd＝1.123；
The SOC of the battery of the electric automobile and the driving mileage d thereof approximately satisfy the following linear relation:
E＝(1-d/D)×100％；
the last trip end time approximately follows the Weibull distribution as follows:
Figure BDA0002532655320000132
in the formula: k is a radical oftIs a shape parameter, ctIs a scale parameter;
setting the electric automobile to be charged only once a day, wherein the charging time starts from the end of the last driving; and the residual electric quantity of the electric automobile user is charged when 20% -50%, and the expected off-grid time of the user is uniformly set to be Te7, expected charge S when user is away from homeeUniform distribution of 80% -100% is obeyed; as shown in fig. 16, it is a flowchart of an algorithm for calculating a disordered charge electric vehicle by using a monte carlo simulation method;
d. the electric automobile schedulable charge-discharge load model:
in the set model, each automobile is only subjected to once discharging scheduling in one day, and schedulable charging and discharging capacity calculation is carried out on the second and third types of electric automobiles:
as shown in FIG. 1, t is set for a time node and an SOC state of a schedulable electric vehicle during a scheduling process0To t1The time period of (1) is the time node length of the SOC of the electric vehicle, wherein t and t are also includedlimThe two nodes, t0The moment when the electric automobile arrives at the charging place and is connected into the power grid is shown, and the electric quantity of the battery of the electric automobile is Q0(ii) a t represents the current time of the electric automobile, and the electric quantity of the battery of the electric automobile is Q at the momentt(ii) a From time t0The state of charge to t is known, and the period of time represents that the electric vehicle is scheduled by the power grid for a period of time; t is t1At the moment when the electric vehicle leaves the power grid, the state of charge of the electric vehicle is Q1(ii) a Then from t to t1The time interval is the time interval of the electric vehicle to be scheduled by the power grid; during the period, the electric automobile is subjected to discharge scheduling, and the discharge process is transferred from t to continue to tlimTime of day tlimThe method comprises the steps of representing a discharge limit time point of the electric automobile, wherein the time point represents the maximum time when the electric automobile is discharged under the condition of meeting the expected state of charge when a user of the electric automobile leaves the network, and if the battery capacity of the electric automobile cannot be full when the electric automobile leaves the network after the time point is exceededSufficient to the user; electric automobile at tlimState of charge at time Qlim(ii) a From tlimTo t1The period of time is a period of time when the electric automobile needs to be charged after being discharged, and because the electric quantity of the battery of the electric automobile is in a lower level through the previous discharging, the electric automobile needs to be charged in the later period of time to meet the electric quantity required by the user when the user is off the network; the flow chart is shown in FIG. 17;
electric automobile discharge electrode time limit tlimThe calculation is as follows:
Figure BDA0002532655320000141
in the formula: t is the current time of the electric automobile; the battery charge quantity of the electric automobile at the time t is Q1；t1The moment when the electric automobile leaves the power grid; pcCharging power for the electric vehicle; pdDischarging power for the electric vehicle; csThe battery capacity of the electric automobile;
the schedulable discharge capacity of the electric automobile is calculated as follows:
Figure BDA0002532655320000142
the electric vehicle charging capacity is calculated as follows:
Pev,c＝Cs[Qt+Q1-Q0-Qlim]
extracting a schedulable electric automobile matrix from the electric automobile state matrix by using a Monte Carlo method, determining the number of schedulable electric automobiles, and determining the discharge limit time point t of each schedulable electric automobilelimThe schedulable charging and discharging capacity is added to the schedulable charging and discharging capacity curve of each schedulable electric automobile, and finally the schedulable electric automobile charging and discharging limit capacity is obtained;
(4) the optimization model established in the day-ahead scheduling stage comprises a target function with the lowest total operation cost of the regional intelligent power grid and a distributed power supply constraint condition;
a. the lowest objective function of the total operation cost of the regional intelligent power grid in the day-ahead scheduling stage takes the fuel cost, the starting and stopping cost of a conventional unit and the electricity purchasing and selling cost of the electric automobile into account, and the expression is as follows:
Figure BDA0002532655320000151
f is the total operation cost of the system in the day-ahead economic optimization scheduling stage; n is a radical of1The number of the thermal power generating units I,
Figure BDA0002532655320000152
a fuel cost function of the unit i in a time period t;
Figure BDA0002532655320000153
generating power of the unit i in a time period t;
Figure BDA0002532655320000154
starting and stopping a unit i at a time t; n is a radical of2The number of the thermal power generating units II is,
Figure BDA0002532655320000155
as a function of the fuel cost for unit i over time t,
Figure BDA0002532655320000156
the generated power of the unit i in the time period t,
Figure BDA0002532655320000157
is the starting and stopping state of the unit i in the time period t,
Figure BDA0002532655320000158
the method comprises the steps of obtaining a start-stop cost function of a unit i in a time period t;
Figure BDA0002532655320000159
scheduling Capacity for charging of a slow-charging schedulable electric vehicle at time t, μcFor the charging price of the electric vehicle in the time period t,
Figure BDA00025326553200001510
scheduling Capacity for discharging of a trickle charge schedulable electric vehicle at time t, μdThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure BDA00025326553200001511
scheduling the capacity of the electric vehicle for fast charging in the t time period;
Figure BDA00025326553200001512
scheduling capacity for discharging of fast-charging schedulable electric vehicle in t period ηbScheduling battery loss penalty cost for the fast-charging schedulable electric vehicle;
Figure BDA00025326553200001513
scheduling capacity for charging of the disordered electric vehicle at a time period t;
b. in order to ensure the safe and reliable operation of the regional intelligent power grid, each unit in the regional intelligent power grid needs to satisfy a certain equality constraint or inequality constraint condition in each time interval, and the method comprises the following steps:
i thermal power generating unit output upper and lower limit constraint and climbing constraint:
Figure BDA00025326553200001514
Figure BDA00025326553200001515
wherein the content of the first and second substances,
Figure BDA0002532655320000161
the minimum output and the maximum output of the unit i are respectively set;
Figure BDA0002532655320000162
respectively limiting the power of climbing up and down slopes of the unit i;
II thermal power generating units output upper and lower limit constraints, start and stop constraints and climbing constraints:
Figure BDA0002532655320000163
Figure BDA0002532655320000164
Figure BDA0002532655320000165
Figure BDA0002532655320000166
wherein the content of the first and second substances,
Figure BDA0002532655320000167
the minimum output and the maximum output of the unit i are respectively set;
Figure BDA0002532655320000168
respectively the continuous on-off time and the minimum on-off time of the unit i;
Figure BDA0002532655320000169
respectively limiting the power of climbing up and down slopes of the unit i;
scheduling capacity constraint of the slow charging electric automobile:
Figure BDA00025326553200001610
Figure BDA00025326553200001611
Figure BDA00025326553200001612
wherein the content of the first and second substances,
Figure BDA00025326553200001613
respectively charging the slow charging electric automobileLimiting;
Figure BDA00025326553200001614
the upper limit of the discharge capacity of the slow charging electric automobile;
fourthly, the schedulable capacity of the fast-charging electric automobile is restrained:
Figure BDA00025326553200001615
Figure BDA00025326553200001616
Figure BDA00025326553200001617
wherein the content of the first and second substances,
Figure BDA00025326553200001618
respectively setting the upper limit and the lower limit of the charging capacity of the fast charging electric automobile;
Figure BDA00025326553200001619
the upper limit of the discharge capacity of the quick-charging electric automobile is defined;
system load balancing constraint:
Figure BDA0002532655320000171
wherein the content of the first and second substances,
Figure BDA0002532655320000172
the wind power plant output power is in a time period t;
Figure BDA0002532655320000173
normal load power for a period t;
(5) according to the objective function with the lowest total operation cost of the regional intelligent power grid in the step (4) and the constraint condition of the distributed power supply; solved by the yalcip software module of MATLAB: the method comprises the following steps that in each period of time in the future, the operation state of a thermal power generating unit I, the operation state of a thermal power generating unit II, the charge and discharge power of a slow-charge schedulable electric automobile and the charge and discharge power of a fast-charge schedulable electric automobile are set; scheduling the total running cost of the system day ahead; see fig. 9, 10, 11 and 12, respectively; the total operation cost of the system predicted by the day-ahead scheduling is 423.72 ten thousand yuan/day;
II, model training stage:
(1) in the model training phase scheduling plan, 15 minutes are taken as unit time intervals, and the whole day is divided into 96 time intervals;
(2) simulating and generating electric field generating power, disordered charging electric vehicle charging power and conventional load fluctuation condition data at each time period in a day by a disturbance increasing method;
(3) combining the electric field generating power, the disordered charging electric vehicle charging power and the conventional load fluctuation condition of each time period in the day generated by simulation with the day-ahead scheduling data to be used as a deep learning network input sample;
(4) in the model training stage, the thermal power generating unit II and the slow-charging schedulable electric automobile execute a day-ahead scheduling plan;
(5) the operation and maintenance cost of the thermal power generating unit I, the operation and maintenance cost of the fast-charging schedulable electric automobile and the operation and maintenance cost of the disordered charging electric automobile are considered by optimizing the objective function in the model training stage, and the expression is as follows:
Figure BDA0002532655320000174
wherein, FXThe total cost of system operation in the model training stage; n is a radical of1The number of the thermal power generating units I,
Figure BDA0002532655320000175
as a function of the fuel cost for unit i over time t,
Figure BDA0002532655320000176
the generated power of the unit i in the time period t,
Figure BDA0002532655320000177
starting and stopping a unit i in a time period tA state;
Figure BDA0002532655320000178
for the charge scheduling capacity of the fast charge schedulable electric automobile in the time period t,
Figure BDA0002532655320000179
capacity of discharge scheduling for fast-charging schedulable electric vehicle at time t, ηb-XScheduling Battery loss penalty cost, μ for fast Charge schedulable electric vehiclescCharging price of electric vehicle in t period, mudThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure BDA00025326553200001710
scheduling capacity for charging of the disordered electric vehicle at a time period t;
(6) in order to ensure safe and reliable operation of the regional intelligent power grid, in a model training phase, all units in the regional intelligent power grid meet the constraint conditions and are the same as those in a day-ahead scheduling phase;
(7) and (3) solving a model training stage through a yalcip software module of MATLAB according to the optimization objective function in the step (5) and the constraint condition in the step (6): the thermal power generating unit I simulates generating power, and the fast charging adjustable electric automobile simulates charging and discharging power; the system operation total cost and the like in the model training stage are obtained, and the solved simulated power generation power of the thermal power generating unit I and the simulated charge and discharge power of the fast-charging schedulable electric automobile are used as output samples of the deep learning network;
a. in the model training and scheduling stage, the input samples in the deep learning network training samples are as follows:
Figure BDA0002532655320000181
in the formula: n is a radical ofinputSimulating input samples for the deep learning network within a time period t; t is a time scale;
Figure BDA0002532655320000182
simulating the difference between the wind power plant power and the forecast power of the wind power plant in the day before in the time period t;
Figure BDA0002532655320000183
simulating the difference between the conventional load and the predicted conventional load in the day ahead for the time period t;
Figure BDA0002532655320000184
the difference between the load of the simulated disordered charging electric vehicle in the time period t and the predicted load of the disordered charging electric vehicle in the day ahead is calculated;
Figure BDA0002532655320000185
simulating wind farm power for a time period t;
Figure BDA0002532655320000186
simulating a conventional load for a time period t;
Figure BDA0002532655320000187
simulating the load of the disordered charging electric vehicle in the time period t;
Figure BDA0002532655320000188
scheduling capacity for discharging of the slow-charging schedulable PEV at a time t;
Figure BDA0002532655320000189
discharging scheduling capacity of the slow charging schedulable PEV in a t period;
Figure BDA00025326553200001810
discharging scheduling capacity of the PEV in the t period is scheduled for the quick charging of the day-ahead scheduling stage;
Figure BDA00025326553200001811
the charging scheduling capacity of the PEV in the t time period can be scheduled for the quick charging in the day-ahead scheduling stage;
Figure BDA00025326553200001812
generating power of a unit i in the thermal power generating unit I in a day-ahead scheduling stage in a t period;
Figure BDA00025326553200001813
generating power of a unit i in a thermal power unit II in a time period t;
Figure BDA00025326553200001814
predicting power for a wind farm day ahead within a time period t;
Figure BDA00025326553200001815
predicting a conventional load for a day ahead within a time period t;
Figure BDA00025326553200001816
the load of the disorderly charged electric automobile is predicted in the day ahead;
b. taking the scheduling data of the controllable units in the model training stage as output samples of deep learning network training, wherein the output samples in the deep learning network training samples are as follows:
Figure BDA00025326553200001817
in the formula: n is a radical ofoutputSimulating output samples for the deep learning network within the time period t;
Figure BDA00025326553200001818
simulating discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA00025326553200001819
the simulation charging power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA00025326553200001820
the method comprises the following steps that (1) the analog output power of a thermal power generating unit I in a time period t is obtained;
(8) repeating the steps (2) to (7), adding input samples and output samples, and training a deep learning network to obtain an intraday scheduling model;
thirdly, scheduling stage in a day:
(1) in the scheduling stage in the day, taking 15 minutes as a unit time interval, dividing the whole day into 96 time intervals;
(2) forecasting the generated power of the wind power plant, the charging power of the disordered electric vehicle and the conventional load fluctuation condition at each time interval in a day in an ultra-short period;
(3) inputting the next-time ultra-short-term prediction data and the day-ahead scheduling plan into a day scheduling model to obtain the power generation power of the thermal power generating unit I and the charge-discharge power of the fast-charging schedulable electric vehicle as next-time scheduling values, taking the day-time ultra-short-term prediction data as actual operation conditions, and setting the actual power of the wind power plant, the actual power of the disordered charge electric vehicle and the conventional load as shown in fig. 4, 6 and 8:
the input data of the deep learning network are as follows:
Figure BDA0002532655320000191
in the formula: n is a radical ofS-inputInputting data in real time for the deep learning network within a time period t;
Figure BDA0002532655320000192
the difference between the ultra-short-term predicted power of the wind power plant in the day and the predicted power of the wind power plant in the day before is the time period t;
Figure BDA0002532655320000193
the difference between the ultra-short-term predicted conventional load in the day and the predicted conventional load before the day within the time interval t is obtained;
Figure BDA0002532655320000194
the difference between the super short-term predicted load of the unordered charging electric automobile in the day and the predicted load of the unordered charging electric automobile in the day in the time period t;
Figure BDA0002532655320000195
ultra-short-term power prediction is carried out on the wind power plant in the day within the time period t;
Figure BDA0002532655320000196
predicting the conventional load for the ultra-short term in the day within the time period t;
Figure BDA0002532655320000197
is disorder within the day within a time period tThe ultra-short-term load prediction of the charged electric automobile.
The output data of the deep learning network is as follows:
Figure BDA0002532655320000198
in the formula: n is a radical ofS-outputReal-time output data of the deep learning network in a representation time period t;
Figure BDA0002532655320000199
representing the real-time discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA00025326553200001910
representing the real-time charging power of the fast-charging schedulable electric automobile in a time period t;
Figure BDA0002532655320000201
and the real-time output power of the thermal power generating unit I in the time period t is shown.
Comparing the scheduling result in the day with the scheduling result of the BP neural network:
as shown in fig. 13 and 14, the BP neural network scheduling cost is 477.43 ten thousand yuan, and the intra-day scheduling cost is 456.25 ten thousand yuan.
(4) And in the day dispatching stage, the thermal power generating unit II and the slow charging dispatchable electric automobile execute a day-ahead dispatching plan.
(5) And the economy of scheduling in the future scheduling verification day is improved, the data of the whole day in the day are collected, and the planning is carried out again in the future. The future scheduling cost is 456.13 ten thousand yuan.
The scope of the invention is not limited to the above embodiments, and various modifications and changes may be made by those skilled in the art, and any modifications, improvements and equivalents within the spirit and principle of the invention should be included in the scope of the invention.

Claims (1)

1. The utility model provides a take into account smart power grids deep learning scheduling method of schedulable electric automobile fast/slow charging and discharging form which characterized in that: the method comprises a day-ahead scheduling stage, a model training stage and a day scheduling stage;
the day-ahead scheduling stage:
(1) segmenting according to hours, dividing 1 day into 24 time periods, and taking the power output and absorption of each distributed unit in each time period as fixed values;
(2) calling the predicted wind power plant generating power, the charging power of the disordered charging electric vehicle and the conventional load fluctuation condition data in each time period in the future day;
(3) establishing a mathematical model of the electric automobile:
a. classifying the electric automobile:
according to the charging characteristics of the electric automobile, the electric automobile is divided into three categories: the first type is an unordered charging electric automobile, the second type is a fast-charging schedulable electric automobile, and the third type is a slow-charging schedulable electric automobile;
b. electric vehicle state matrix:
the state of any electric automobile at the end of driving is represented by a one-dimensional matrix:
Ω＝[L N SnSeTsTe]
wherein: l represents the type of the electric automobile load, and numbers 1, 2, 3, 4 and 5 respectively represent the disordered charging load, the slow-charging schedulable discharging load, the fast-charging schedulable charging load and the fast-charging schedulable discharging load of the electric automobile; n represents a charging and discharging identifier of the electric automobile, the charging mode is 1, the discharging mode is-1, and the rest moments are 0; snAnd SeRespectively representing the state of charge of the electric automobile when the electric automobile is stopped and the expected state of charge of a user when the electric automobile is off the network; t issAnd TeRespectively representing the network access time of the electric automobile and the network leaving time of a user;
c. unordered electric automobile load model:
setting: the residual electric quantity of the electric automobile user is charged when being 20% -50%, and the expected off-grid time of the user is uniformly set to be Te7, expected charge S when user is away from homeeUniform distribution of 80-100% is obeyed; electric driving by adopting Monte Carol method of Monte CarolExtracting the final driving end time of the electric automobile, the charge state when the electric automobile finishes driving and the daily driving mileage of an electric automobile user in the automobile state matrix to generate a disordered electric automobile model:
d. the electric automobile schedulable charge-discharge load model:
in the set model, each automobile is only subjected to once discharging scheduling in one day, and schedulable charging and discharging capacity calculation is carried out on the second and third types of electric automobiles:
setting t0To t1The time period of (1) is the time node length of the SOC of the electric vehicle, wherein t and t are also includedlimThe two nodes, t0The moment when the electric automobile arrives at the charging place and is connected into the power grid is shown, and the electric quantity of the battery of the electric automobile is Q0(ii) a t represents the current time of the electric automobile, and the electric quantity of the battery of the electric automobile is Q at the momentt(ii) a From time t0The state of charge to t is known, and the period of time represents that the electric vehicle is scheduled by the power grid for a period of time; t is t1At the moment when the electric vehicle leaves the power grid, the state of charge of the electric vehicle is Q1(ii) a Then from t to t1The time interval is the time interval of the electric vehicle to be scheduled by the power grid; during the period, the electric automobile is subjected to discharge scheduling, and the discharge process is transferred from t to continue to tlimTime of day tlimThe method comprises the steps that a discharge limit time point of the electric automobile is shown, the time point shows the maximum discharging moment of the electric automobile under the condition that the expected state of charge when a user of the electric automobile leaves the network is met, and if the battery capacity of the electric automobile when the electric automobile leaves the network exceeds the time point, the battery capacity cannot meet the user requirement; electric automobile at tlimState of charge at time Qlim(ii) a From tlimTo t1The period of time is a period of time when the electric automobile needs to be charged after being discharged, and because the electric quantity of the battery of the electric automobile is in a lower level through the previous discharging, the electric automobile needs to be charged in the later period of time to meet the electric quantity required by the user when the user is off the network;
electric automobile discharge electrode time limit tlimThe calculation is as follows:
Figure FDA0002532655310000021
in the formula: t is the current time of the electric automobile; the battery charge quantity of the electric automobile at the time t is Q1；t1The moment when the electric automobile leaves the power grid; pcCharging power for the electric vehicle; pdDischarging power for the electric vehicle; csThe battery capacity of the electric automobile;
the schedulable discharge capacity of the electric automobile is calculated as follows:
Figure FDA0002532655310000022
the electric vehicle charging capacity is calculated as follows:
Pev,c＝Cs[Qt+Q1-Q0-Qlim]
extracting a schedulable electric automobile matrix from the electric automobile state matrix by using a Monte Carlo method, determining the number of schedulable electric automobiles, and determining the discharge limit time point t of each schedulable electric automobilelimThe schedulable charging and discharging capacity is added to the schedulable charging and discharging capacity curve of each schedulable electric automobile, and finally the schedulable electric automobile charging and discharging limit capacity is obtained;
(4) the optimization model established in the day-ahead scheduling stage comprises a target function with the lowest total operation cost of the regional intelligent power grid and a distributed power supply constraint condition;
a. the lowest objective function of the total operation cost of the regional intelligent power grid in the day-ahead scheduling stage takes the fuel cost, the starting and stopping cost of a conventional unit and the electricity purchasing and selling cost of the electric automobile into account, and the expression is as follows:
Figure FDA0002532655310000031
f is the total operation cost of the system in the day-ahead economic optimization scheduling stage; n is a radical of1The number of the thermal power generating units I,
Figure FDA0002532655310000032
a fuel cost function of the unit i in a time period t;
Figure FDA0002532655310000033
generating power of the unit i in a time period t;
Figure FDA0002532655310000034
starting and stopping a unit i at a time t; n is a radical of2The number of the thermal power generating units II is,
Figure FDA0002532655310000035
as a function of the fuel cost for unit i over time t,
Figure FDA0002532655310000036
the generated power of the unit i in the time period t,
Figure FDA0002532655310000037
is the starting and stopping state of the unit i in the time period t,
Figure FDA00025326553100000314
the method comprises the steps of obtaining a start-stop cost function of a unit i in a time period t;
Figure FDA0002532655310000038
scheduling Capacity for charging of a slow-charging schedulable electric vehicle at time t, μcFor the charging price of the electric vehicle in the time period t,
Figure FDA0002532655310000039
scheduling Capacity for discharging of a trickle charge schedulable electric vehicle at time t, μdThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure FDA00025326553100000310
for fast-charging and scheduling electric vehicle at time tA charge scheduling capacity;
Figure FDA00025326553100000311
scheduling capacity for discharging of fast-charging schedulable electric vehicle in t period ηbScheduling battery loss penalty cost for the fast-charging schedulable electric vehicle;
Figure FDA00025326553100000312
scheduling capacity for charging of the disordered electric vehicle at a time period t;
b. in order to ensure the safe and reliable operation of the regional intelligent power grid, each unit in the regional intelligent power grid needs to satisfy a certain equality constraint or inequality constraint condition in each time interval, and the method comprises the following steps:
i thermal power generating unit output upper and lower limit constraint and climbing constraint:
Figure FDA00025326553100000313
Figure FDA0002532655310000041
wherein the content of the first and second substances,
Figure FDA0002532655310000042
the minimum output and the maximum output of the unit i are respectively set;
Figure FDA0002532655310000043
respectively limiting the power of climbing up and down slopes of the unit i;
II thermal power generating units output upper and lower limit constraints, start and stop constraints and climbing constraints:
Figure FDA0002532655310000044
Figure FDA00025326553100000419
Figure FDA00025326553100000420
Figure FDA0002532655310000045
wherein the content of the first and second substances,
Figure FDA0002532655310000046
respectively the minimum and maximum output of unit i βi ont、βi offt、
Figure FDA0002532655310000047
Respectively the continuous on-off time and the minimum on-off time of the unit i;
Figure FDA0002532655310000048
respectively limiting the power of climbing up and down slopes of the unit i;
scheduling capacity constraint of the slow charging electric automobile:
Figure FDA0002532655310000049
Figure FDA00025326553100000410
Figure FDA00025326553100000411
wherein the content of the first and second substances,
Figure FDA00025326553100000412
respectively is the upper limit and the lower limit of the charging capacity of the slow charging electric automobile;
Figure FDA00025326553100000413
the upper limit of the discharge capacity of the slow charging electric automobile;
fourthly, the schedulable capacity of the fast-charging electric automobile is restrained:
Figure FDA00025326553100000414
Figure FDA00025326553100000415
Figure FDA00025326553100000416
wherein the content of the first and second substances,
Figure FDA00025326553100000417
respectively setting the upper limit and the lower limit of the charging capacity of the fast charging electric automobile;
Figure FDA00025326553100000418
the upper limit of the discharge capacity of the quick-charging electric automobile is defined;
system load balancing constraint:
Figure FDA0002532655310000051
wherein the content of the first and second substances,
Figure FDA0002532655310000052
the wind power plant output power is in a time period t;
Figure FDA0002532655310000053
normal load power for a period t;
(5) according to the objective function with the lowest total operation cost of the regional intelligent power grid in the step (4) and the constraint condition of the distributed power supply; solved by the yalcip software module of MATLAB: the method comprises the following steps that in each period of time in the future, the operation state of a thermal power generating unit I, the operation state of a thermal power generating unit II, the charge and discharge power of a slow-charge schedulable electric automobile and the charge and discharge power of a fast-charge schedulable electric automobile are set; scheduling the total running cost of the system day ahead;
II, model training stage:
(1) in the model training phase scheduling plan, 15 minutes are taken as unit time intervals, and the whole day is divided into 96 time intervals;
(2) simulating and generating electric field generating power, disordered charging electric vehicle charging power and conventional load fluctuation condition data at each time period in a day by a disturbance increasing method;
(3) combining the electric field generating power, the disordered charging electric vehicle charging power and the conventional load fluctuation condition of each time period in the day generated by simulation with the day-ahead scheduling data to be used as a deep learning network input sample;
(4) in the model training stage, the thermal power generating unit II and the slow-charging schedulable electric automobile execute a day-ahead scheduling plan;
(5) the operation and maintenance cost of the thermal power generating unit I, the operation and maintenance cost of the fast-charging schedulable electric automobile and the operation and maintenance cost of the disordered charging electric automobile are considered by optimizing the objective function in the model training stage, and the expression is as follows:
Figure FDA0002532655310000054
wherein, FXThe total cost of system operation in the model training stage; n is a radical of1The number of the thermal power generating units I,
Figure FDA0002532655310000055
as a function of the fuel cost for unit i over time t,
Figure FDA0002532655310000056
the generated power of the unit i in the time period t,
Figure FDA0002532655310000057
starting and stopping a unit i at a time t;
Figure FDA0002532655310000058
for the charge scheduling capacity of the fast charge schedulable electric automobile in the time period t,
Figure FDA0002532655310000059
capacity of discharge scheduling for fast-charging schedulable electric vehicle at time t, ηb-XScheduling Battery loss penalty cost, μ for fast Charge schedulable electric vehiclescCharging price of electric vehicle in t period, mudThe discharge electricity price of the electric automobile in the time period t is obtained;
Figure FDA00025326553100000510
scheduling capacity for charging of the disordered electric vehicle at a time period t;
(6) in order to ensure safe and reliable operation of the regional intelligent power grid, in a model training phase, all units in the regional intelligent power grid meet the constraint conditions and are the same as those in a day-ahead scheduling phase;
(7) and (3) solving a model training stage through a yalcip software module of MATLAB according to the optimization objective function in the step (5) and the constraint condition in the step (6): the thermal power generating unit I simulates generating power, and the fast charging adjustable electric automobile simulates charging and discharging power; the system operation total cost and the like in the model training stage are obtained, and the solved simulated power generation power of the thermal power generating unit I and the simulated charge and discharge power of the fast-charging schedulable electric automobile are used as output samples of the deep learning network;
a. in the model training and scheduling stage, the input samples in the deep learning network training samples are as follows:
Figure FDA0002532655310000061
in the formula: n is a radical ofinputSimulating input samples for the deep learning network within a time period t; t is a time scale;
Figure FDA0002532655310000062
simulating the difference between the wind power plant power and the forecast power of the wind power plant in the day before in the time period t;
Figure FDA0002532655310000063
simulating the difference between the conventional load and the predicted conventional load in the day ahead for the time period t;
Figure FDA0002532655310000064
the difference between the load of the simulated disordered charging electric vehicle in the time period t and the predicted load of the disordered charging electric vehicle in the day ahead is calculated;
Figure FDA0002532655310000065
simulating wind farm power for a time period t;
Figure FDA0002532655310000066
simulating a conventional load for a time period t;
Figure FDA0002532655310000067
simulating the load of the disordered charging electric vehicle in the time period t;
Figure FDA0002532655310000068
scheduling capacity for discharging of the slow-charging schedulable PEV at a time t;
Figure FDA0002532655310000069
discharging scheduling capacity of the slow charging schedulable PEV in a t period;
Figure FDA00025326553100000610
discharging scheduling capacity of the PEV in the t period is scheduled for the quick charging of the day-ahead scheduling stage;
Figure FDA00025326553100000611
the charging scheduling capacity of the PEV in the t time period can be scheduled for the quick charging in the day-ahead scheduling stage;
Figure FDA00025326553100000612
generating power of a unit i in the thermal power generating unit I in a day-ahead scheduling stage in a t period;
Figure FDA00025326553100000613
generating power of a unit i in a thermal power unit II in a time period t;
Figure FDA00025326553100000614
predicting power for a wind farm day ahead within a time period t;
Figure FDA00025326553100000615
predicting a conventional load for a day ahead within a time period t;
Figure FDA00025326553100000616
the load of the disorderly charged electric automobile is predicted in the day ahead;
b. taking the scheduling data of the controllable units in the model training stage as output samples of deep learning network training, wherein the output samples in the deep learning network training samples are as follows:
Figure FDA00025326553100000617
in the formula: n is a radical ofoutputSimulating output samples for the deep learning network within the time period t;
Figure FDA00025326553100000618
simulating discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure FDA00025326553100000619
the simulation charging power of the fast-charging schedulable electric automobile in a time period t;
Figure FDA00025326553100000620
the method comprises the following steps that (1) the analog output power of a thermal power generating unit I in a time period t is obtained;
(8) repeating the steps (2) to (7), adding input samples and output samples, and training a deep learning network to obtain an intraday scheduling model;
thirdly, scheduling stage in a day:
(1) in the scheduling stage in the day, taking 15 minutes as a unit time interval, dividing the whole day into 96 time intervals;
(2) forecasting the generated power of the wind power plant, the charging power of the disordered electric vehicle and the conventional load fluctuation condition at each time interval in a day in an ultra-short period;
(3) inputting the next-time ultra-short term prediction data and a day-ahead scheduling plan into a day scheduling model to obtain the power generation power of the thermal power generating unit I and the charge-discharge power of the fast-charging schedulable electric automobile as next-time scheduling values;
the input data of the deep learning network are as follows:
Figure FDA0002532655310000071
in the formula: n is a radical ofS-inputInputting data in real time for the deep learning network within a time period t;
Figure FDA0002532655310000072
the difference between the ultra-short-term predicted power of the wind power plant in the day and the predicted power of the wind power plant in the day before is the time period t;
Figure FDA0002532655310000073
the difference between the ultra-short-term predicted conventional load in the day and the predicted conventional load before the day within the time interval t is obtained;
Figure FDA0002532655310000074
the difference between the predicted load of the ultra-short period disordered charging electric vehicle in the day and the predicted load of the disordered charging electric vehicle in the day before is obtained within the time period t;
Figure FDA0002532655310000075
ultra-short-term power prediction is carried out on the wind power plant in the day within the time period t;
Figure FDA0002532655310000076
predicting the conventional load for the ultra-short term in the day within the time period t;
Figure FDA0002532655310000077
predicting the load of the disorderly charged electric automobile in the time period t within the ultra-short term;
the output data of the deep learning network is as follows:
Figure FDA0002532655310000078
in the formula: n is a radical ofS-outputReal-time output data of the deep learning network in a representation time period t;
Figure FDA0002532655310000079
representing the real-time discharge power of the fast-charging schedulable electric automobile in a time period t;
Figure FDA00025326553100000710
representing the real-time charging power of the fast-charging schedulable electric automobile in a time period t;
Figure FDA00025326553100000711
representing the real-time output power of the thermal power generating unit I in a time period t;
(4) and in the day dispatching stage, the thermal power generating unit II and the slow charging dispatchable electric automobile execute a day-ahead dispatching plan.
CN202010522503.0A 2020-06-10 2020-06-10 Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle Active CN111626527B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010522503.0A CN111626527B (en) 2020-06-10 2020-06-10 Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010522503.0A CN111626527B (en) 2020-06-10 2020-06-10 Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle

Publications (2)

Publication Number Publication Date
CN111626527A true CN111626527A (en) 2020-09-04
CN111626527B CN111626527B (en) 2023-02-03

Family

ID=72261067

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010522503.0A Active CN111626527B (en) 2020-06-10 2020-06-10 Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle

Country Status (1)

Country Link
CN (1) CN111626527B (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112134304A (en) * 2020-09-22 2020-12-25 南方电网数字电网研究院有限公司 Micro-grid full-automatic navigation method, system and device based on deep learning
CN112241596A (en) * 2020-10-26 2021-01-19 国网上海市电力公司 Urban power grid medium-long term production simulation method considering electric automobile
CN112396217A (en) * 2020-11-02 2021-02-23 国家电网有限公司华东分部 Controllable load resource long-distance optimal scheduling implementation method and system
CN112488444A (en) * 2020-10-31 2021-03-12 海南电网有限责任公司 Electric vehicle fast-slow synchronous ordered charging scheduling method and electric quantity settlement method
CN112803493A (en) * 2021-02-09 2021-05-14 国网山东省电力公司电力科学研究院 Marginal cost-based distributed energy storage and distributed photovoltaic cooperation method and system
CN112952839A (en) * 2021-01-29 2021-06-11 国网内蒙古东部电力有限公司 Power distribution network economic dispatching evaluation method based on controllable load
CN113780670A (en) * 2021-09-16 2021-12-10 太原理工大学 Two-stage-based regional power grid electric vehicle peak shaving optimization scheduling method

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120229089A1 (en) * 2011-03-08 2012-09-13 Trilliant Networks, Inc. System and Method For Managing Load Distribution Across a Power Grid
CN105930955A (en) * 2016-04-07 2016-09-07 浙江万马新能源有限公司 Deep learning-based charging network operation situation analysis method and apparatus
CN106505552A (en) * 2016-11-10 2017-03-15 太原理工大学 A kind of double-deck bus direct-current grid and its control method based on power pond
CN106712061A (en) * 2016-05-16 2017-05-24 浙江工业大学 Intra-day priority scheduling method based on electric-vehicle schedulable capability
CN109995091A (en) * 2019-04-26 2019-07-09 太原理工大学 A kind of alternating current-direct current mixing micro-capacitance sensor economic load dispatching method considering prediction error
CN110570014A (en) * 2019-08-07 2019-12-13 浙江大学 Electric vehicle charging load prediction method based on Monte Carlo and deep learning
CN111047071A (en) * 2019-10-29 2020-04-21 国网江苏省电力有限公司盐城供电分公司 Power system real-time supply and demand interaction method based on deep migration learning and Stackelberg game
CN111064214A (en) * 2019-12-31 2020-04-24 国网浙江省电力有限公司电力科学研究院 Power distribution network optimal scheduling method based on electric vehicle two-stage rolling strategy

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20120229089A1 (en) * 2011-03-08 2012-09-13 Trilliant Networks, Inc. System and Method For Managing Load Distribution Across a Power Grid
CN105930955A (en) * 2016-04-07 2016-09-07 浙江万马新能源有限公司 Deep learning-based charging network operation situation analysis method and apparatus
CN106712061A (en) * 2016-05-16 2017-05-24 浙江工业大学 Intra-day priority scheduling method based on electric-vehicle schedulable capability
CN106505552A (en) * 2016-11-10 2017-03-15 太原理工大学 A kind of double-deck bus direct-current grid and its control method based on power pond
CN109995091A (en) * 2019-04-26 2019-07-09 太原理工大学 A kind of alternating current-direct current mixing micro-capacitance sensor economic load dispatching method considering prediction error
CN110570014A (en) * 2019-08-07 2019-12-13 浙江大学 Electric vehicle charging load prediction method based on Monte Carlo and deep learning
CN111047071A (en) * 2019-10-29 2020-04-21 国网江苏省电力有限公司盐城供电分公司 Power system real-time supply and demand interaction method based on deep migration learning and Stackelberg game
CN111064214A (en) * 2019-12-31 2020-04-24 国网浙江省电力有限公司电力科学研究院 Power distribution network optimal scheduling method based on electric vehicle two-stage rolling strategy

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
A.A.S.MOHAMED等: "Predictive active power-flow control of two-way wireless power transfer system in V2G services", 《2016 IEEE 2ND ANNUAL SOUTHERN POWER ELECTRONICS CONFERENCE》 *
师景佳 等: "计及电动汽车可调度能力的风/车协同参与机组组合策略", 《高压电技术》 *

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112134304A (en) * 2020-09-22 2020-12-25 南方电网数字电网研究院有限公司 Micro-grid full-automatic navigation method, system and device based on deep learning
CN112241596A (en) * 2020-10-26 2021-01-19 国网上海市电力公司 Urban power grid medium-long term production simulation method considering electric automobile
CN112488444A (en) * 2020-10-31 2021-03-12 海南电网有限责任公司 Electric vehicle fast-slow synchronous ordered charging scheduling method and electric quantity settlement method
CN112488444B (en) * 2020-10-31 2022-12-23 海南电网有限责任公司 Electric vehicle fast and slow synchronous ordered charging scheduling method and electric quantity settlement method
CN112396217A (en) * 2020-11-02 2021-02-23 国家电网有限公司华东分部 Controllable load resource long-distance optimal scheduling implementation method and system
CN112396217B (en) * 2020-11-02 2022-05-06 国家电网有限公司华东分部 Controllable load resource long-distance optimal scheduling implementation method and system
CN112952839A (en) * 2021-01-29 2021-06-11 国网内蒙古东部电力有限公司 Power distribution network economic dispatching evaluation method based on controllable load
CN112952839B (en) * 2021-01-29 2022-09-20 国网内蒙古东部电力有限公司 Power distribution network economic dispatching evaluation method based on controllable load
CN112803493A (en) * 2021-02-09 2021-05-14 国网山东省电力公司电力科学研究院 Marginal cost-based distributed energy storage and distributed photovoltaic cooperation method and system
CN113780670A (en) * 2021-09-16 2021-12-10 太原理工大学 Two-stage-based regional power grid electric vehicle peak shaving optimization scheduling method
CN113780670B (en) * 2021-09-16 2023-08-15 太原理工大学 Two-stage-based regional power grid electric automobile peak shaving optimization scheduling method

Also Published As

Publication number Publication date
CN111626527B (en) 2023-02-03

Similar Documents

Publication Publication Date Title
CN111626527B (en) Intelligent power grid deep learning scheduling method considering fast/slow charging/discharging form of schedulable electric vehicle
CN105024432B (en) A kind of electric automobile discharge and recharge Optimization Scheduling based on virtual electricity price
CN112186809B (en) Virtual power plant optimization cooperative scheduling method based on V2G mode of electric vehicle
CN109217290B (en) Microgrid energy optimization management method considering electric vehicle charging and discharging
CN109449973B (en) Energy optimization method for rail transit power supply system containing photovoltaic and energy storage
CN109948823B (en) Self-adaptive robust day-ahead optimization scheduling method for light storage charging tower
CN110138006B (en) Multi-microgrid coordinated optimization scheduling method considering new energy electric vehicle
CN110378548B (en) Electric automobile virtual power plant multi-time scale response capability assessment model construction method
Alabi et al. Improved hybrid inexact optimal scheduling of virtual powerplant (VPP) for zero-carbon multi-energy system (ZCMES) incorporating Electric Vehicle (EV) multi-flexible approach
CN103793758A (en) Multi-objective optimization scheduling method for electric vehicle charging station including photovoltaic power generation system
CN112103946A (en) Particle swarm algorithm-based microgrid energy storage optimization configuration method
Kheradmand-Khanekehdani et al. Well-being analysis of distribution network in the presence of electric vehicles
CN103246942A (en) Multi-objective scheduling method for wind power-electric automobile-thermal power combined operation model
CN107919675B (en) Charging station load scheduling model comprehensively considering benefits of vehicle owners and operators
CN103904749B (en) A kind ofly consider the orderly charge control method of the electric automobile of wind power output fluctuation
CN107590607A (en) A kind of micro-capacitance sensor Optimal Scheduling and method based on photovoltaic prediction
CN112070628B (en) Multi-target economic dispatching method for smart power grid considering environmental factors
CN112238781A (en) Electric automobile ordered charging control method based on layered architecture
CN113326467B (en) Multi-target optimization method, storage medium and optimization system for multi-station fusion comprehensive energy system based on multiple uncertainties
CN109038655B (en) Method for calculating matched energy storage capacity of large photovoltaic power station under power limiting requirement
Li et al. Optimal dispatch for PV-assisted charging station of electric vehicles
CN111762057B (en) Intelligent charging and discharging management method for V2G electric vehicle in regional microgrid
CN113131529A (en) Renewable energy bearing capacity assessment method considering multiple flexible resources
CN111224418B (en) Micro-grid scheduling method and system based on electric vehicle energy storage
CN115940284B (en) Operation control strategy of new energy hydrogen production system considering time-of-use electricity price

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