Method and system for hierarchical scheduling of energy flow in microgrid for deploying battery energy storage systems

Abstract

Methods and systems manage energy flow between an electrical grid and a plurality of microgrids that deploy at least one battery energy storage system and are controlled by a microgrid controller. A central controller receives grid information from an external source and is arranged to generate an aggregated long-term schedule and an aggregated short-term schedule of energy flow and energy price for a plurality of microgrids connected with the central controller. A microgrid scheduler associated with each microgrid includes a long-term day-ahead scheduler that receives an aggregated long-term schedule and a short-term near-real-time scheduler that receives an aggregated short-term schedule. The long-term day-ahead scheduler is arranged to generate a microgrid long-term schedule that controls energy flow between the grid and each microgrid and its connected battery energy storage system, the microgrid long-term schedule being corrected by a microgrid short-term near-real-time schedule generated by the microgrid short-term near-real-time scheduler.

Description

Technical Field

[0001] The present disclosure generally relates to control of a grid-connected microgrid deploying a battery energy storage system. More specifically, the present disclosure relates to methods and systems for controlling energy flow between a grid and a microgrid from a battery energy storage system using hierarchical dispatch. Background Art

[0002] Currently, most electricity is generated by large centralized power plants, such as nuclear power plants, hydroelectric power plants, and fossil fuel power plants. These large facilities often use non-renewable energy sources to generate electricity, such as coal or natural gas. Such power plants usually have good economies of scale, but due to various economic and operational reasons, they may not be able to provide all the electricity required to serve the dynamically changing loads of the power grid. For example, due to the imbalance of supply and demand in the power grid and the environmental impact caused by seasonal effects, the load is increased at certain times of the day and the aging centralized power plants cause power outages. A battery energy storage system (BESS) that stores the generated power can be connected to a power plant, a substation, a transmission line, or a customer site to selectively use the stored energy to supplement or provide all the electricity required for the microgrid, thereby preventing service interruptions.

[0003] The BESS employs chemical energy storage batteries that store energy chemically, such as, for example, lithium-ion (LiON) batteries, lead-acid batteries (Pb), or sodium-sulfur (NAS) batteries. Considering the power capacity of the power conversion system (PCS), the batteries included in the BESS are charged using the PCS attached to the BESS. The power from the PCS for charging the BESS batteries may come from renewable energy sources such as solar or wind power or from non-renewable sources such as power from a diesel generator or provided from the grid.

[0004] The power and energy capacity of a BESS is determined by the number of battery racks contained in the BESS and put into use. When multiple BESS containers are deployed to be attached to a single or one or more PCSs, a controller associated with each BESS can calculate a power baseline for all PCSs. To cope with future peak power demands, the BESS is required to be fully charged before the peak demand arrives. Charging at high power at an earlier time can result in the peak being shifted only to an earlier time, thereby avoiding co-occurrence peaks in the grid. However, time shifting cannot avoid the demand charge that would be generated when charging the BESS with the same amount of electricity. Avoiding high demand charge requires not only shifting the BESS charging to another time, but also dispersing the charging over a longer duration, which results in a reduction in the power drawn from the grid source at any given time. Charging over a longer duration requires scheduling the charging operation in advance so that charging can start in advance. Summary of the invention

[0005] The present disclosure relates to methods and systems for controlling a microgrid connected to a power grid and deploying a battery energy storage system.

[0006] In a first embodiment, a method for managing energy flow between a power grid and a plurality of microgrids is disclosed, wherein the plurality of microgrids deploy at least one battery energy storage system controlled by a microgrid controller. The method includes receiving power grid information from an external source by a central controller, and generating an aggregated long-term schedule and an aggregated short-term schedule for energy flow and energy price of a plurality of microgrids connected to the central controller by the central controller using the power grid information. The method also includes: receiving the aggregated long-term schedule and the aggregated short-term schedule by a microgrid scheduler, the microgrid scheduler using the aggregated long-term scheduler to generate a microgrid long-term schedule for controlling energy flow between the power grid and each microgrid and its connected battery energy storage system, wherein the microgrid long-term schedule is corrected using the microgrid short-term near real-time schedule generated by the microgrid scheduler.

[0007] In a second embodiment, a system for managing energy flow between a power grid and a plurality of microgrids is disclosed, each microgrid deploying at least one battery energy storage system and being controlled by a microgrid controller. The system includes a central controller arranged to receive power grid information from an external source and arranged to generate an aggregated long-term schedule and an aggregated short-term schedule for energy flow and energy prices for a plurality of microgrids connected to the central controller. A microgrid scheduler is associated with each microgrid connected to the central controller, the microgrid scheduler including a long-term day-ahead scheduler receiving the aggregated long-term schedule and a short-term near-real-time scheduler receiving the aggregated short-term schedule. The long-term day-ahead scheduler is arranged to generate a microgrid long-term schedule for controlling energy flow between the power grid and each microgrid and its connected battery energy storage system, the microgrid long-term schedule being corrected by a microgrid short-term near-real-time schedule generated by the microgrid short-term near-real-time scheduler.

[0008] Other technical features will be apparent to those skilled in the art from the following drawings, descriptions and claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] For a more complete understanding of the present disclosure, reference is now made to the following description taken in conjunction with the accompanying drawings, in which:

[0010] Figure 1 is a diagram schematically illustrating a hierarchical control system for aggregation of microgrids connected to a power grid and controlled by a virtual power plant controller according to an exemplary embodiment of the present disclosure, wherein each microgrid deploys a battery energy storage system;

[0011] Figure 2is a diagram schematically illustrating a communication network and features of a battery energy storage system according to an exemplary embodiment of the present disclosure;

[0012] Figure 3 is a diagram schematically illustrating a microgrid ECS controller according to an exemplary embodiment of the present disclosure;

[0013] Figure 4 is a diagram schematically illustrating a VPP controller according to an exemplary embodiment of the present disclosure;

[0014] Figure 5 is a diagram schematically illustrating a scheduler for microgrid aggregation according to an exemplary embodiment of the present disclosure;

[0015] Figure 6 is a diagram schematically illustrating a scheduler for generating a scheduler model of a microgrid according to an exemplary embodiment of the present disclosure;

[0016] Figure 7 is a schematic diagram showing an exemplary embodiment according to the present disclosure Figure 5 a diagram of the functional elements of a scheduler; and

[0017] Figure 8 is a schematic diagram showing an exemplary embodiment according to the present disclosure Figure 6 Diagram of the functional elements of a scheduler. DETAILED DESCRIPTION

[0018] These figures discussed below and the various embodiments for describing the principles of the present invention in this patent document are carried out by way of example only and should not be construed as limiting the scope of the present invention in any way. Those skilled in the art will appreciate that the principles of the present invention can be implemented in any type of suitably arranged device or system.

[0019] Figure 1A hierarchical control system 100 for controlling an aggregation of microgrids connected to a grid and controlled by a virtual power plant (VPP) controller is shown, each microgrid deploying one or more battery energy storage systems (BESS) consisting of four hierarchical levels. The system 100 uses hierarchical scheduling for controlling the energy flow between the grid and the connected microgrids, which store energy in the battery energy storage system at certain times and dispatch the stored energy from the battery energy storage system at various times. At the first level, a BESS unit controller 110 is located in a BESS container 145. The BESS unit controller 110 is used to control the functions of the BESS container 145 and its power conversion system (PCS). Each BESS container 145 is organized as an independent package that may include at least a power conversion system, a battery system, a heating ventilation and air conditioning (HVAC) system, a fire protection system, and components and sensors required to monitor the BESS container 145. Each BESS container 145 can be used to power an independent deployment of BESS, such as, for example, a building or a commercial enterprise or a microgrid deployment of a single or multiple BESS containers 145 to provide power to a residential or commercial area.

[0020] At the second level of the hierarchical control system 100, an energy control system (ECS) 130 is communicatively coupled to one or more BESS unit controllers 110, 110'. The ECS 130 includes an ECS controller 135 that controls the operation of one or more BESS containers 145. The ECS 130 may be connected to a single standalone BESS container 145, to a multiple BESS container 145 deployment, or to a grid-connected microgrid 120 having a multiple container BESS deployment. For example, in Figure 1 , the ECS 130 is shown connected to the BESS unit controller 110 and the BESS unit controller 110 ′ of the BESS container 145 ′ via a communication line 109 .

[0021] The task of the BESS unit controller 110, 110' is to provide safe and reliable operation of the BESS container 145. The BESS unit controller 110 monitors the operation of the BESS container 145, thereby preventing operation during fault conditions, shutting down faulty subsystems and / or sending notifications and alarms to the operator station 164 or mobile device 240. If a component, sensor or subsystem of the BESS container 145 fails or malfunctions, alarms may be sent using different priority levels. The BESS unit controller 110 interfaces with all subsystems within the container, such as, for example, the power conversion system, battery system, HVAC subsystem, fire protection system, etc. The main components of the BESS container 145 include the BESS unit controller 110, the battery rack housed within the BESS container 145, and a power container (not shown) that houses the PCS on a separate removable skid.

[0022] Figure 2 The communication features of the BESS container 145 are schematically shown. Each BESS container 145 includes a communication interface that operates using, for example, an Ethernet communication protocol to communicate between the various components of the BESS container 145, and to communicate to other BESS containers that are communicatively coupled to each other in a multi-container deployment, such as, for example, between BESS containers 145 and 145'. Each microgrid 120 is connected to the VPP 160 via a network switch 225, a firewall 220, and a network 108, as well as to a server 162, an operator station 164, and a cloud 168. The communication between the BESS container 145 and the VPP 160 is based on the MQ Telemetry Transport (MQTT), or Distributed Network (DNP3) protocol used in supervisory control and data acquisition (SCADA) and remote monitoring systems. The network switching device 215 contained in the BESS container 145 distributes control and data signals between the components of the BESS container 145, and distributes the control and data signals to other interconnected BESS containers, such as, for example, the BESS container 145'. For example, a network switch 215' located in a BESS container 145' may receive control and data signals from an ECS 130 contained in a microgrid 120. The network switch 225 distributes the control and data signals from the ECS 130 to the BESS unit controller 110' via the network switch 215' using the communication network 109. The BESS container 145 may also include a communication interface 210 that may be connected to a wireless interface 230, thereby allowing communication with the BESS unit controller 110 using a handheld mobile device 240. The handheld mobile device 240 may be used to configure operating parameters of the BESS container 145, and / or retrieve operating data, fault conditions, and / or alarms and warnings from the BESS unit controller 110. The BESS container 145 may also include other forms of communication interfaces, such as, for example, serial digital and analog I / O communication interfaces, which allow the BESS unit controller 110 to communicate and control the BESS container 145 subsystems.

[0023] Re-reference Figure 1 , a MECS controller 140 is communicatively coupled to the ECS controller 135. The MECS controller 140 comprises a third level of the BESS control system 100. The MECS controller 140 manages alternative power generation assets, such as, for example, solar, wind, hydroelectric power, which are connectable to and available on the grid for use by the BESS container 145. The MECS controller 140 is arranged to control and manage the use of alternative power sources for a standalone BESS container 145 deployment or for a plurality of microgrid connected BESS container deployments.

[0024] The ECS controller 135 transmits the status of a fleet of BESS containers 145 to the MECS controller 140. The ECS controller 135 also transmits the power reference of the PCS to the BESS unit controllers 110, 110' within the BESS container 145. The MECS controller 140 receives a schedule autonomously generated by the VPP controller 160 located in the remote operation center. The schedule may include: a schedule for power exchange between the microgrid and the grid; a power price schedule that may be associated with the import and export of power between the microgrid and the grid; and a schedule for use cases to be put into the BESS container and the generation assets and loads. For example, renewable energy smoothing exported between 09:00 and 12:00 in a day, frequency regulation support between 12:00 and 17:00, and grid peak demand support between 17:00 and 20:00, etc.

[0025] The MECS controller 140 calculates a schedule for charging or discharging one or more BESS containers 145 by taking into account the schedule received from the VPP 160 for grid power exchange and local generation and load within the microgrid. The construction of the schedule by the MECS controller 140 involves predictive operations to predict local generation and demand / load. When the power generation is controllable, such as when using, for example, a diesel generator, the forecast will be an agreed schedule for power from the controllable generator. When the power generation is uncontrollable and intermittent in nature, such as from a renewable energy generator, the forecast is constructed using historical time series data and inputs from meteorological stations and other weather forecast sources. Such calculated forecasts have uncertainty, especially at the end of the forecast period. This can result in a deviation from the forecast and the actual, which needs to be taken into account by calculating a revised schedule for charging or discharging the batteries of the BESS container 145.

[0026] When calculating the predictive schedule, the MECS controller 140 uses short-term battery charge / discharge forecasts because the charging / discharging needs of the battery depend on the finite amount of time of the battery's charge or C-Rating. For example, it takes 1 hour to fully charge a 1C-rated system, while it takes 4 hours to fully charge or discharge a 0.25C-rated system. Therefore, for a 0.25C-rated system, in order to meet the grid peak support use case at any given time, battery charging should have started at least four hours before the time the battery is scheduled to be discharged. Even for a 1C-rated system where charging with high power is feasible, charging with a higher power may attract a higher demand charge, or alternatively may result in a larger increase in battery temperature, which may cause the battery to degrade faster. A lower charging power will mean that the charging will last longer, and at least a forecast of this duration is necessary to ensure that the BESS battery is charged when its stored energy needs to be dispatched.

[0027] The MECS controller 140 is also used to calculate a forecast of power demand within the microgrid based on the connected controllable loads. The scheduling for power demand is subject to time or production constraints, such as HVAC loads and uncontrollable loads, where the power demand may deviate from the forecasted demand. When there is a deviation from the forecasted generation or demand in generation or demand, or both, a revised schedule that takes into account the deviation and the current state needs to be calculated.

[0028] The VPP 160 comprises the fourth level of the BESS control system 100. A software operating program executed on, for example, a server 162 serves as a central controller that controls the functions of the VPP 160. The VPP 160 also includes an operator station 164 and an interface to a cloud 168. The server 162 may be any device that provides resources, data, services, or software programs to other processing devices or clients via a communication network. The operator station 164 may be any computing device that provides functionality for power plant operation and monitoring, including displaying graphics such as diagrams, systems, BESS container 145 deployments, and data to a user or operator. The operator station 164 may also receive input from a user or operator to adjust or input configurable parameters for the BESS unit controller 110, the BESS ECS controller 135, and / or the MECS controller 140. The cloud 168 may be any computing device or technology that provides services over the Internet, including information, data storage, servers, access to databases, networking, and software. The VPP 160 may control multiple BESS containers 145 connected to the VPP 160 via a communication network 108. As Figure 1 The VPP 160 shown controls a multi-microgrid 120 deployment, such as Figure 1 BESS containers 145 , 145 ′ in microgrid 1 , microgrid 2 , and microgrid N are shown.

[0029] Figure 3 An exemplary schematic diagram of a MECS controller 140 of an ECS 130 is shown. The MECS controller 140 includes a computer processor 310, a memory 315, and a hardware infrastructure 320. The hardware infrastructure 320 provides I / O and network circuitry for communicatively connecting the MECS controller 140 to the power generation, operator input, and battery storage system of the VPP controller 160 and its associated BESS. The processor 310 executes host operating system software 325 used in running the MECS controller. The memory 315 can be any form of known memory device for storing data, a host operating system, and other application software that provides specialized functionality, such as, for example, application-level virtualization software that allows the container engine 330 to execute the microgrid predictor model 370 and the scheduler model 360 as runtime applications in separate computing containers 380, 385.

[0030] Based on inputs from the microgrid, each MECS controller 140 generates a forecast of local demand and local power generation for its associated microgrid. These inputs include, for example, instantaneous microgrid power demand, microgrid power generation, battery state of charge (SOC) from a BESS battery storage system connected to the microgrid, real-time prices local to the microgrid, and local weather station data at the microgrid site served by the MECS controller 140. MECS140 uses microgrid inputs to generate demand forecasts for local power demand for microgrids controlled by ECS130. MECS controller 140 transmits the state of the microgrid to server 162, where an operator using operator station 164 performs manual scheduling using a human-machine interface. Alternatively, a VPP controller 160 operating in server 162 can autonomously generate scheduling for grid power exchange for different microgrids controlled by server 162 or VPP controller 160.

[0031] Figure 4 An exemplary schematic diagram of a VPP controller 160 is shown. The VPP controller 160 autonomously calculates a VPP schedule for grid power exchange for all microgrids connected to the VPP controller 160 by utilizing state information received from each MECS controller in the MECS controller 140. The VPP schedule involves a predictive operation that requires the use of a forecast of the availability of power from renewable energy generating sources located within the microgrid and other energy generating sources that provide their energy to the grid but are located outside the microgrid. In calculating the VPP schedule, the VPP schedule also uses a forecast of grid power demand. The VPP controller 160 is also configured to receive information from an external source, such as, for example, energy costs from an energy trading platform, to consider the cost of purchasing energy and the revenue from selling energy in the calculation of the grid power exchange schedule.

[0032] The VPP schedule calculated by the VPP controller 160 is an aggregate schedule for all microgrids connected to the VPP 160. The VPP controller 160 uses information received from an external source 400 to generate a grid demand forecast 420, a grid generation forecast 425, and a grid price forecast 430 based on historical data provided to the VPP controller 160 from a utility grid operator 418, an independent system operator (ISO) 415, an energy trading platform 408, or an energy market operator 405, a carbon trading platform 412, or a carbon market operator 410. For example, the energy market operator 405 may receive information on the settlement price of the local grid from the energy trading platform 408, and provide the trading platform 408 with real-time sales prices, real-time purchase prices, and forecasts of sales prices and purchase prices for electricity supplied by the grid, which trades between users of the grid on the trading platform to buy or sell electrical energy. The carbon market operator 410 may receive carbon bid prices and carbon clearing prices that provide carbon price requirements for the grid from the carbon trading platform 412.

[0033] The VPP controller 160 also generates a microgrid demand forecast 460 and a microgrid generation forecast 465 using information provided by each microgrid connected to the VPP controller 160. The microgrid demand forecast 460 uses the microgrid information provided by each MECS controller 140 and may include the actual demand of the loads in each microgrid, the generation capacity of each microgrid, the current state of charge (SOC) of the BESS supporting each microgrid, the capacity and response time of the BESS supporting each microgrid, and the weather forecast for the location of the microgrid. The BESS response time varies from microgrid to microgrid, depending on the SOC of the BESS batteries and their availability to provide power. The microgrid generation forecast 465 receives the forecast of generation and demand generated by the microgrid predictor model 370 of each MECS controller 140 from each microgrid connected to the VPP controller 160.

[0034] The VPP 160 controller uses the aggregated energy and price scheduler 450 to generate a schedule of aggregated energy flows and energy prices for the microgrids connected to the VPP 160 controller using data from the grid forecasts 420, 425 and 430 and the microgrid forecasts 460, 465. The VPP controller 160 takes into account the cost of energy imported from external centralized generation sources available at different times of the day and when the power demand and energy cost from centralized generation are high, allowing the VPP controller 160 to provide a price schedule for sharing energy between microgrids by generating a local price schedule for selling energy and buying energy for the microgrids. The price schedule is calculated so that the price schedule enables each microgrid to sell excess generation or stored energy at a higher price, buy energy when the price is low, or limit load and use stored energy from the BESS when the load cannot be interrupted. The price schedule also enables excess generation to be stored at the BESS for subsequent sale at a higher price, rather than limiting generation.

[0035] The aggregated schedule generated by the aggregated energy and price scheduler 450 is used to generate a schedule of grid power exchanges and power and energy prices for each individual microgrid connected to the VPP 160 controller. The aggregated energy and price schedule from the scheduler 450 is input to the individual energy and price schedulers 452, 454, 456 of the microgrid. Each microgrid energy and price scheduler 452, 454, 456 also receives microgrid state information from the MECS controller scheduler model 360 from its associated MECS controller 140. Each energy and price scheduler 452, 454, and 456 uses the microgrid state information and the aggregated energy and price schedule to generate a long-term schedule and a short-term scheduler model for each microgrid connected to the VPP controller 160.

[0036] The aggregated schedule for multiple microgrids generated by the aggregated energy and price scheduler 450 is used to generate a schedule for grid power exchange, power and energy prices for each microgrid controlled by the VPP. The aggregated schedule is input to the energy and price schedulers 452, 454, 456 of each microgrid, and includes a long-term scheduler and a short-term scheduler for each individual microgrid. The long-term schedule for energy and price is downloaded to the scheduler model 360 of each MECS controller 140 associated with the microgrid. The MECS controller 140 runs a time series prediction model to generate a forecast of local power generation and demand using a microgrid predictor model 370. The MECS controller 140 also runs a scheduler model 360 for generating a short-term near real-time schedule to schedule power flows between the microgrid and the main grid. The short-term scheduler model is trained using data generated from its corresponding microgrid short-term schedule from the VPP controller 160. The trained scheduler model is then downloaded from the cloud, from each individual energy and price scheduler 452, 454 and 456 to its associated MECS controller 140.

[0037] Figure 5 An example VPP scheduler 500 is schematically shown. The VPP scheduler includes a long-term day-ahead scheduler. Those skilled in the art will well understand that the scheduler can be configured to generate a schedule for more than one day in advance. The long-term day-ahead scheduler 510 and the VPP short-term near real-time scheduler 520 are executed in the energy price scheduler 450 of the VPP controller 160. The VPP long-term day-ahead scheduler 510 generates an energy purchase and sale schedule for the microgrid within the VPP, which takes into account the demand and generation forecasts of the microgrid. The input 511 of the VPP long-term day-ahead scheduler 510 includes the long-term grid power, sales, purchase and demand forecasts provided by the grid forecasts 420, 425 and 430 and the aggregated demand, generation and aggregated initial battery system state of charge forecasts from the microgrid demand and generation forecasts 460, 465.

[0038] The output 512 of the VPP long-term day-ahead scheduler 510 is a long-term dispatch power demand forecast 512, which is output from the long-term scheduler 510 and is also applied as an input to the VPP short-term near real-time scheduler 520 along with inputs 521 for short-term forecasts of grid prices, purchases, sales and microgrids, demand, and generation. The VPP short-term inputs 521 take into account the real-time response time of the BESS battery state of charge and the forecast of energy availability based on the battery cycles scheduled when generating a short-term schedule 522 for aggregated power imports or exports for the microgrid. The short-term real-time scheduler inputs 521 correct the long-term schedule generated by the VPP long-term scheduler 510 to correct for any inaccuracies due to uncertainties in the forecasts generated for weather, grid power purchase and sales prices, grid power demand, carbon prices, and aggregated power demand and generation from the microgrid. The VPP short-term near real-time scheduler 520 outputs a short-term schedule 522 for aggregated power imports / exports for the microgrid, which is input to each microgrid energy and price scheduler 452, 454, 456.

[0039] Figure 6 An example of a microgrid scheduler 600 is schematically shown. The microgrid scheduler 600 consists of a long-term day-ahead scheduler 610 and a short-term near real-time scheduler 620. Each microgrid scheduler 600 is executed within a separate energy and price scheduler 452, 454, and 456 associated with microgrid 1, microgrid 2, and microgrid N, respectively. The input 611 of the long-term day-ahead scheduler 610 includes at least long-term microgrid demand and generation forecasts, long-term energy price forecasts, local weather forecasts, and long-term dispatch power demand forecast output 512 from the VPP long-term scheduler 510. The long-term day-ahead scheduler 610 outputs a schedule 612 consisting of a long-term battery charging / discharging schedule, a long-term load dispatch schedule, and a long-term controllable generation schedule. The schedule output 612 is also applied as an input to the short-term near real-time scheduler 620 together with the short-term forecast input 621. The input 621 provides short-term microgrid demand, generation, and battery state of charge forecasts. The input 621 also includes short-term weather station and real-time data and real-time response time forecasts for the BESS battery state of charge.

[0040] The short-term real-time scheduler 620 corrects the long-term schedule generated by the microgrid long-term day-ahead scheduler 610 to correct for any inaccuracies due to uncertainties in the forecasts generated for weather, grid generation, load dispatch, and battery charging / discharging. The microgrid short-term real-time scheduler 620 generates a short-term battery charging / discharging schedule 622, a short-term load dispatch schedule 623, and a short-term controllable generation schedule 624 output from the MECS controller 140 associated with each microgrid energy and price scheduler 452, 454, 456.

[0041] The hierarchical scheduling of the present disclosure uses the higher level scheduling of the VPP long-term and short-term scheduler 500 for the aggregation of microgrids to minimize grid power import by enabling energy sharing between microgrids controlled by VPPs. Its mission is also to minimize the cost of grid power import and maximize the revenue from selling energy from VPPs to external energy trading partners.

[0042] The objective of the long-term day-ahead scheduler 610 of each microgrid energy and price scheduler 452 , 454 , 456 is to minimize the energy cost of the respective microgrid while satisfying the constraints that the VPP controller 160 has in satisfying external grid power imports and exports.

[0043] Figure 7 An example of functional operating elements of a VPP scheduler 500 for generating an aggregated long-term schedule 512 and a short-term schedule 522 for a microgrid is schematically shown. The long-term day-ahead scheduler 510 generates the long-term schedule 512 using a second instance of a mixed integer nonlinear programming (MINLP) solver 551. The first instance of the MINLP solver 550 receives and uses Figure 5 The long-term input data 511 listed in , with long-term predicted generation and demand data, price costs of the grid, and aggregation of microgrids as constraints, provides an optimization objective for minimizing the cost of energy purchases and maximizing revenue from energy sales. The generated long-term schedule 512 is typically a day-ahead schedule, but may extend over a longer time frame depending on the energy market mechanism. The forecast uncertainty estimator 558 estimates a measure of uncertainty in the forecast by measuring the difference between past day-ahead forecasts provided by the historical database 560 using current energy prices and past actual generation demand. The estimated uncertainty is taken into account in the MINLP 551 when generating the aggregated long-term schedule 512 for the microgrid.

[0044] The long-term schedule 512 is generated once a day. The delay in calculating the long-term schedule 512 does not significantly affect the operation of the aggregation of microgrids. In order to minimize the waiting time for generating the short-term schedule, a neural network scheduler 552 with a recurrent neural network (RNN) model 553 and a fast Fourier transform (FFT) preprocessor 554 is used to learn and simulate the operation of the MINLP solver 551. The performance of the neural network scheduler 552 is continuously measured by calculating the total cost of energy based on the scheduled grid exchange power flow between the aggregation of microgrids and the grid. The cost is calculated based on the schedule generated by the RNN model 553 and compared with the total cost associated with the ideal schedule generated by the first instance of the MINLP solver 550. The scheduler performance analyzer 570 receives the actual cost and revenue input from the energy power demand and carbon cost algorithm 565 and the ideal cost and revenue generated by the second energy power demand and carbon cost algorithm 566 that receives the ideal schedule from the MINLP solver 550. The scheduler performance analyzer 570 uses the actual and ideal cost and revenue inputs to generate a value that is a function of the difference between the total cost associated with the actual short-term schedule generated by the neural network scheduler 552 of the scheduler 520 and the ideal schedule generated by the MINLP solver 550 using data from the historical database 560. The value from the scheduler performance analyzer 570 is fed back into the policy update algorithm 575, which continuously updates the RNN 553, thereby implementing the policy function for the VPP short-term near real-time scheduler 520.

[0045] Figure 8 An example of functional operating elements of a microgrid scheduler 600 for generating a long-term schedule 612 and a short-term schedule 622, 623, 624 for each microgrid is schematically shown. The long-term day-ahead scheduler 610 generates a long-term schedule 612 for battery energy storage and dispatch, power generation from controllable generators, and controllable loads using a second example mixed integer nonlinear programming (MINLP) solver 651. The MINLP solver 651 receives and uses Figure 6 The optimization objective is to minimize the total cost of operation, including battery life cycle cost, and maximize the revenue from energy sales, using the long-term input data listed in 611, as well as demand and energy prices and forecasts within the VPP and manual scheduling as constraints.

[0046] The long-term schedule 612 generated for each microgrid is typically a day-ahead schedule, but may extend over a slightly longer time frame depending on the energy market mechanism. The forecast uncertainty estimator 658 estimates the uncertainty in the forecast by measuring the difference between past forecasts and past actual power generation provided by the historical database 660 using the demand and price of energy. The second instance of the MINLP solver 651 takes this uncertainty into account when generating the long-term schedule 612 for each microgrid. Since the long-term schedule is generated once a day, the delay in calculating the schedule for battery energy storage and dispatch, power generation from controllable generators, and controllable loads will not significantly affect the operation of the microgrid. However, the generation of short-term near real-time schedules 622, 623, 624 is accompanied by waiting time constraints. The delay in calculating short-term near real-time schedules 622, 623, 624 may potentially affect microgrid operation.

[0047] To minimize the waiting time for generating a short-term schedule, a neural network scheduler 652 with a recurrent neural network (RNN) model 653 and a fast Fourier transform (FFT) preprocessor 654 is used to learn and simulate the operation of the first instance MINLP solver 650. The performance of the neural network scheduler 652 is continuously measured by calculating the total cost of energy based on the scheduled controllable generation, load dispatch, and battery charging / discharging. The cost is calculated based on the schedule generated by the RNN model 653 and compared to the total cost associated with the ideal schedule generated by the first instance MINLP solver 650. The scheduler performance analyzer 670 receives the actual cost and revenue input from the energy power demand and carbon cost algorithm 665 and the ideal cost and revenue generated by the second energy power demand and carbon cost algorithm 666 that receives the ideal schedule from the MINLP solver 650. The scheduler performance analyzer 670 uses the actual and ideal cost and revenue inputs to generate a value that is a function of the difference between the total cost associated with the actual short-term schedule generated by the RNN 652 of the scheduler 620 and the ideal schedule generated by the first instance MINLP solver 650 using data from the historical database 560. The performance analyzer 670 includes an anomaly detection algorithm that is used to identify situations where the performance difference is due to an abnormal reason, which should not be used to update the RNN 653 model policy. For normal situations, the values ​​from the scheduler performance analyzer 670 are fed back into the policy update algorithm 675, which continuously updates the RNN 653, thereby implementing the policy function of the short-term near real-time scheduler 620.

[0048] It may be advantageous to set forth the definitions of certain words and phrases used throughout this patent document. The term "communication" and its derivatives cover both direct communication and indirect communication. The terms "include" and "comprising" and their derivatives mean including but not limited to this. The term "or" is inclusive, meaning and / or. The phrase "associated with..." and its derivatives may mean including, included within, interconnected with, included within, connected to, or connected with, coupled to, or coupled with, can communicate with, collaborate with, interlace, juxtapose, approach, be coupled to, or be combined with, have, have the property of, have a relationship with, or have a relationship with. When used with a list of items, the phrase "at least one of..." means that different combinations of one or more items in the listed items may be used, and only one item in the list may be needed. For example, "at least one of A, B, and C" includes any combination of A, B, C, A and B, A and C, B and C, and A and B and C.

[0049] The description in this application should not be construed as implying that any particular element, step, or function is an essential or critical element that must be included within the scope of the claims. The scope of subject matter protected by a patent is limited only by the claims as allowed. In addition, none of the claims is intended to invoke 35 U.S.C. §112(f) with respect to any of the appended claims or claim elements, unless the exact words "means for..." or "step for..." followed by a participle phrase identifying the function are expressly used in a particular claim. The use of terms such as (but not limited to) "mechanism", "module", "device", "unit", "component", "element", "member", "device", "machine", "system" or "controller" within the claims is understood and intended to refer to structures known to those skilled in the relevant art, as further modified or enhanced by the features of the claims themselves, and is not intended to invoke 35 U.S.C. §112(f).

[0050] Although the present disclosure has described certain embodiments and generally associated methods, changes and permutations of these embodiments and methods will be apparent to those skilled in the art. Therefore, the above description of the exemplary embodiments does not limit or restrict the present disclosure. Other changes, substitutions and variations are also possible without departing from the spirit and scope of the present disclosure as defined in the following claims.

Claims

1. A method for managing energy flow between a power grid and a plurality of microgrids, each microgrid deploying at least one battery energy storage system and controlled by a microgrid controller, the method comprising: Receiving grid information from an external source by a central controller; generating, by the central controller, an aggregated long-term schedule and an aggregated short-term schedule for energy flows and energy prices of the plurality of microgrids connected to the central controller using the grid information; The microgrid scheduler receives the aggregated long-term schedule and the aggregated short-term scheduler; generating, by each microgrid scheduler, a microgrid long-term schedule using the aggregated long-term scheduler, the microgrid long-term schedule being used to control energy flow between the grid and each microgrid and its connected battery energy storage system; as well as The microgrid short-term near real-time schedule generated by the microgrid scheduler is used to correct each microgrid long-term schedule.

2. The method according to claim 1, wherein the method further comprises: training a microgrid controller dispatcher model at the central controller; as well as The microgrid controller scheduler model is downloaded to each microgrid controller.

3. The method according to claim 1, wherein the microgrid long-term scheduling for controlling energy flow of each microgrid scheduler comprises: Generates long-term battery energy storage system charge / discharge schedules; Generate long-term load dispatch schedules; as well as Generate long-term controllable generation dispatch.

4. The method of claim 1, wherein the microgrid short-term near real-time scheduling corrects the microgrid long-term scheduling for the effects of uncertainty in the microgrid long-term forecast.

5. The method according to claim 4, wherein the method further comprises: In view of the influence of uncertainty in the long-term prediction of the microgrid using the short-term prediction, the microgrid short-term near real-time scheduling is used to correct the long-term scheduling of the microgrid, including at least: electricity generated from renewable energy generators; The battery energy storage system state of charge; a response time of the electricity produced by the renewable energy generator and the battery energy storage system; and Manually entered long-term schedule.

6. The method of claim 5, wherein the microgrid long-term schedule is continuously updated by the microgrid short-term near real-time scheduler based on the short-term forecast and compared with an ideal schedule generated by the microgrid scheduler using data from a historical database.

7. The method according to claim 6, wherein the method further comprises: improving the microgrid scheduler by analyzing a performance difference between the microgrid short-term near real-time schedule and the ideal schedule to generate a value for improving the microgrid scheduler; as well as The improved microgrid dispatcher is downloaded to the microgrid controller, wherein the microgrid controller is used to control the energy flow between the grid and the microgrid to store energy in the microgrid battery energy storage system and to dispatch energy from the microgrid battery energy storage system to the grid and the microgrid.

8. A system for managing energy flow between a power grid and a plurality of microgrids, each microgrid deploying at least one battery energy storage system and controlled by a microgrid controller, the system comprising: a central controller arranged to receive grid information from an external source and arranged to generate an aggregated long-term schedule and an aggregated short-term schedule for energy flows and energy prices for the plurality of microgrids connected to the central controller; a microgrid scheduler associated with each microgrid connected to the central controller, the microgrid scheduler comprising a long-term day-ahead scheduler receiving the aggregated long-term schedule and a short-term near real-time scheduler receiving the aggregated short-term schedule, The long-term day-ahead scheduler is arranged to generate a microgrid long-term schedule for controlling energy flow between the grid and each microgrid and its connected battery energy storage system, and the microgrid long-term schedule is corrected by a microgrid short-term near real-time schedule generated by the microgrid short-term near real-time scheduler.

9. The system of claim 8, wherein the system further comprises a forecast uncertainty estimator operating in the microgrid long-term day-ahead scheduler, the forecast uncertainty estimator estimating a measure of uncertainty from a microgrid long-term forecast to correct the microgrid long-term schedule for the effects of uncertainty in the microgrid long-term forecast, comprising at least: electricity generated from renewable energy generators; The battery energy storage system state of charge; response time of the electricity generated by the renewable energy generator and the battery energy storage system; and Manually entered long-term schedule.

10. The system of claim 9, wherein the system further comprises: a first mixed integer nonlinear programming MINLP solver connected to a historical database, wherein the second MINLP solver is arranged to generate an ideal schedule; a second (MINLP) solver associated with the long-term day-ahead scheduler and connected to the forecast uncertainty estimator, wherein uncertainty estimates for the microgrid long-term forecast are used to generate the microgrid long-term schedule; and a scheduler performance analyzer that receives the ideal schedule from the first MINLP solver and that uses the short term near real time schedule with actual and ideal cost revenue inputs to generate a value that is a function of a difference between a total cost associated with an actual short term near real time schedule generated by the microgrid short term near real time scheduler and the ideal schedule generated by the first MINLP solver, The short-term near real-time scheduler includes a recurrent neural network (RNN) that receives the value generated by the scheduler performance analyzer as input to continuously improve the microgrid scheduler.

Images