WO2017186178A1 - 储能电站自适应动态规划的控制方法、***和存储介质 - Google Patents

储能电站自适应动态规划的控制方法、***和存储介质 Download PDF

Info

Publication number
WO2017186178A1
WO2017186178A1 PCT/CN2017/082564 CN2017082564W WO2017186178A1 WO 2017186178 A1 WO2017186178 A1 WO 2017186178A1 CN 2017082564 W CN2017082564 W CN 2017082564W WO 2017186178 A1 WO2017186178 A1 WO 2017186178A1
Authority
WO
WIPO (PCT)
Prior art keywords
energy storage
module
wind power
control
power
Prior art date
Application number
PCT/CN2017/082564
Other languages
English (en)
French (fr)
Inventor
李相俊
何宇婷
张晶琼
惠东
贾学翠
Original Assignee
中国电力科学研究院
国家电网公司
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 中国电力科学研究院, 国家电网公司 filed Critical 中国电力科学研究院
Priority to US16/095,844 priority Critical patent/US11326579B2/en
Publication of WO2017186178A1 publication Critical patent/WO2017186178A1/zh

Links

Images

Classifications

    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
    • F03D—WIND MOTORS
    • F03D9/00—Adaptations of wind motors for special use; Combinations of wind motors with apparatus driven thereby; Wind motors specially adapted for installation in particular locations
    • F03D9/10—Combinations of wind motors with apparatus storing energy
    • F03D9/11—Combinations of wind motors with apparatus storing energy storing electrical energy
    • H02J3/386—
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
    • F03D—WIND MOTORS
    • F03D7/00—Controlling wind motors 
    • F03D7/02—Controlling wind motors  the wind motors having rotation axis substantially parallel to the air flow entering the rotor
    • F03D7/04—Automatic control; Regulation
    • F03D7/042—Automatic control; Regulation by means of an electrical or electronic controller
    • F03D7/043—Automatic control; Regulation by means of an electrical or electronic controller characterised by the type of control logic
    • F03D7/046—Automatic control; Regulation by means of an electrical or electronic controller characterised by the type of control logic with learning or adaptive control, e.g. self-tuning, fuzzy logic or neural network
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F03—MACHINES OR ENGINES FOR LIQUIDS; WIND, SPRING, OR WEIGHT MOTORS; PRODUCING MECHANICAL POWER OR A REACTIVE PROPULSIVE THRUST, NOT OTHERWISE PROVIDED FOR
    • F03D—WIND MOTORS
    • F03D9/00—Adaptations of wind motors for special use; Combinations of wind motors with apparatus driven thereby; Wind motors specially adapted for installation in particular locations
    • F03D9/20—Wind motors characterised by the driven apparatus
    • F03D9/25—Wind motors characterised by the driven apparatus the apparatus being an electrical generator
    • F03D9/255—Wind motors characterised by the driven apparatus the apparatus being an electrical generator connected to electrical distribution networks; Arrangements therefor
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
    • G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
    • G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
    • G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/004—Generation forecast, e.g. methods or systems for forecasting future energy generation
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28—Arrangements for balancing of the load in a network by storage of energy
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28—Arrangements for balancing of the load in a network by storage of energy
    • H02J3/32—Arrangements for balancing of the load in a network by storage of energy using batteries with converting means
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/381—Dispersed generators
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
    • H02J7/0047—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries with monitoring or indicating devices or circuits
    • H02J7/0048—Detection of remaining charge capacity or state of charge [SOC]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
    • H02J7/007—Regulation of charging or discharging current or voltage
    • H02J7/00712—Regulation of charging or discharging current or voltage the cycle being controlled or terminated in response to electric parameters
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J7/00—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries
    • H02J7/32—Circuit arrangements for charging or depolarising batteries or for supplying loads from batteries for charging batteries from a charging set comprising a non-electric prime mover rotating at constant speed
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2270/00—Control
    • F05B2270/10—Purpose of the control system
    • F05B2270/103—Purpose of the control system to affect the output of the engine
    • F05B2270/1033—Power (if explicitly mentioned)
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2270/00—Control
    • F05B2270/30—Control parameters, e.g. input parameters
    • F05B2270/322—Control parameters, e.g. input parameters the detection or prediction of a wind gust
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2270/00—Control
    • F05B2270/40—Type of control system
    • F05B2270/404—Type of control system active, predictive, or anticipative
    • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
    • F05—INDEXING SCHEMES RELATING TO ENGINES OR PUMPS IN VARIOUS SUBCLASSES OF CLASSES F01-F04
    • F05B—INDEXING SCHEME RELATING TO WIND, SPRING, WEIGHT, INERTIA OR LIKE MOTORS, TO MACHINES OR ENGINES FOR LIQUIDS COVERED BY SUBCLASSES F03B, F03D AND F03G
    • F05B2270/00—Control
    • F05B2270/70—Type of control algorithm
    • F05B2270/709—Type of control algorithm with neural networks
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2300/00—Systems for supplying or distributing electric power characterised by decentralized, dispersed, or local generation
    • H02J2300/20—The dispersed energy generation being of renewable origin
    • H02J2300/28—The renewable source being wind energy
    • 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
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/70—Wind energy
    • Y02E10/72—Wind turbines with rotation axis in wind direction
    • 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
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/70—Wind energy
    • Y02E10/76—Power conversion electric or electronic aspects
    • 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
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E70/00—Other energy conversion or management systems reducing GHG emissions
    • Y02E70/30—Systems combining energy storage with energy generation of non-fossil origin
    • 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

Definitions

  • the invention belongs to the fields of smart grid, energy internet and energy storage technology, and particularly relates to a control method, system and storage medium for adaptive dynamic programming of large-scale energy storage power stations.
  • energy storage can be divided into physical energy storage, electrochemical energy storage and electromagnetic energy storage.
  • battery energy storage is currently in a rapid development stage, and the scale of energy storage power stations has reached megawatts to tens of megawatts. Therefore, it is possible to smooth the power generation of new energy by optimizing the charge and discharge of the battery energy storage system according to the situation of new energy generation by a large-scale battery energy storage system equipped with a certain capacity.
  • the overall optimization is carried out to meet the requirements of the grid for the integration of wind power, photovoltaic power generation and other new energy generation.
  • the traditional first-order low-pass filtering or variable time constant (T) output filtering method due to the inherent delay of the method itself, sometimes control sensitivity Not good.
  • the target value of the energy storage output output by the general moving average filtering algorithm is also largely limited by the input of new energy generation power such as wind power and photovoltaic power generation.
  • new energy generation power such as wind power and photovoltaic power generation.
  • These existing methods encounter sudden changes in the output of new energy power generation.
  • the filtering performance is degraded and affects the subsequent filtering effect.
  • the traditional control method in the adaptive control of the overall output of the energy storage power station its self-learning intelligent optimization control ability needs to be further improved.
  • Embodiments of the present invention provide a control method, system, and storage medium for adaptive dynamic programming of a large-scale energy storage power station, and are expected to reduce the impact of wind power connected to the power grid, and optimize the working capacity and life of the energy storage system. Protection to improve the technical and economical performance of energy storage systems.
  • a control method for adaptive dynamic programming of a large-scale energy storage power plant comprising the following steps:
  • the adaptive dynamic programming control system comprises a two-layer structure of an evaluation module and an execution module, each module is constructed by using a three-layer neural network structure;
  • the control target parameters include Wind power generation capacity Energy storage system capacity W bat , state of charge SOC limit of energy storage system, sampling time ⁇ t, observation time T, volatility control target And volatility limit targets
  • the initialization parameter includes an initial rate of wind power fluctuation
  • the initial value The initial value, the initial value of the state of charge of the energy storage system, and the actual output power of the wind power at the current time.
  • the step (3) includes the following steps:
  • Step 3-1 the original wind power fluctuation rate of the current time t
  • the formula is as follows:
  • I the wind power capacity, that is, the rated power, with It is the maximum and minimum value of wind power during observation time, ⁇ t is the sampling time, T is the observation time, P wp (t) is the original value of wind power, f wp is the original function of wind power fluctuation rate, n is the observation The number of sampling points in time;
  • Step 3-2 smoothing the original value of the wind power according to the rate of change control method
  • the calculation method of the change rate control method is as follows:
  • the rate of change k(t) of wind power is defined according to the original wind power value P wp (t), the smoothed wind power value P hybrid (t), and the sampling time ⁇ t:
  • P wp (t) is the original value of wind power
  • P hybrid (t) is the smoothed wind power value
  • Step 3-3 The wind power fluctuation rate after the change rate control strategy is smoothed
  • the formula is:
  • the goal of the rate of change control power volatility method is to control the smoothed wind power volatility to be less than a given target value during the observation time:
  • Step 3-4 The calculation formula of the energy storage system power is:
  • W bat is the capacity of the energy storage system.
  • the initial training evaluation module and the execution module include: an initial setting discount factor ⁇ , an execution module learning rate l a , an evaluation module learning rate l c , an execution module weight W a , and Evaluating the module weight W c , the maximum number of cycles, and the expected error of the training module, wherein the initial value of the weight is set to a random value between (-1, +1), and the learning rate and the discount factor are selected according to the control effect requirement and Adjustment.
  • the step (5) includes the following steps:
  • Step 5-1 Determine the fluctuation rate of wind storage power Whether it is in the constraint If the energy storage system does not operate, the energy storage power is not corrected; otherwise, the next adaptive dynamic programming module training is performed to seek the optimal energy storage power correction value;
  • Step 5-2 using the controlled object state and the control strategy as input of the evaluation module, training the evaluation module, updating the weight of the evaluation module, and outputting a cost function;
  • Step 5-3 The state of the controlled object includes the wind power fluctuation rate And the stored energy P BESS (t) as the input of the execution module, the training execution module, the weight of the execution module is updated, and the output is the control strategy, that is, the correction value of the stored energy ⁇ P BESS (t);
  • the determining the wind power fluctuation rate is as follows:
  • step 5-2 includes the following steps:
  • Step 5-2-1 the state of the controlled object, that is, the wind power fluctuation rate And normalizing the control strategy, that is, the stored energy correction value ⁇ P BESS (t) to [-1, +1];
  • Step 5-2-2 the state of the controlled object, that is, the wind power fluctuation rate
  • the control strategy that is, the stored energy correction value ⁇ P BESS (t) is sent to the evaluation module as an input, and the output J c WPBESS (t) cost function of the evaluation module is calculated, and the objective function E chybrid (t) is constructed.
  • the evaluation module updates the weight of the evaluation module neural network according to the weight update formula of the evaluation module, and the calculation formula is as follows:
  • the cost function J chybrid (t) is the output of the evaluation module, and the utility function U(t) is about
  • the function is defined according to the control target, ⁇ c is the discount factor;
  • Step 5-2-3 The training of the evaluation module uses a gradient descent method or a particle swarm optimization algorithm to minimize the objective function E chybrid (t) as a target to update the weight W c of the evaluation module neural network, when the objective function
  • the training is completed when E chybrid (t) is reduced to the set error value or the number of iterations reaches the maximum.
  • step 5-3 includes the following steps:
  • Step 5-3-1 the state of the controlled object includes the wind power fluctuation rate And the stored energy P BESS (t) as an input to the execution module, training the execution module;
  • Step 5-3-2 adjust the control strategy, that is, the stored energy correction value ⁇ P BESS (t) by minimizing the evaluation module output J chybrid (t), and the formula is as follows:
  • control strategy ⁇ P BESS (t) is the output of the execution module, used to adjust the stored energy to vary within a reasonable range to reduce the SOC fluctuation range of the energy storage system, u denotes the control strategy ⁇ P BESS (t) is about The function;
  • Step 5-3-3 the training of the execution module uses a gradient descent method or a particle swarm optimization algorithm to update the weight W a of the execution module neural network with the goal of minimizing the objective function E ahybrid (t), when the objective function E ahybrid (t) or an error value is reduced to a set number of iterations reaches the maximum, the training is completed.
  • a control system for adaptive dynamic programming of a large-scale energy storage power plant including parameter initialization module, data acquisition and calculation module, execution module, evaluation module and output module;
  • the parameter initialization module is configured to set a structure and a control target parameter of the adaptive dynamic programming control system, and transmit the target parameter to the data collection and calculation module;
  • the data collection and calculation module is configured to calculate a wind storage power fluctuation rate according to the target parameter, and transmit the wind storage power fluctuation rate to the execution module and the evaluation module;
  • the execution module is configured to obtain a stored energy correction value according to the wind storage power fluctuation rate, and transmit the stored energy correction value to the data acquisition and calculation module, the evaluation module, and the output mode;
  • the evaluation module is configured to derive a cost function according to the wind power fluctuation rate and transmit the cost function to the execution module;
  • the output module is configured to output the control strategy at each moment, the smoothed wind storage power fluctuation rate, the energy storage power, and the energy storage system state of charge SOC.
  • the embodiment of the invention further provides another control method for adaptive dynamic programming of a large-scale energy storage power plant, the method comprising the following steps:
  • the current execution module is based on the An output control strategy, wherein the control strategy is used to control charging and discharging power of an energy storage system of the large-scale energy storage power station;
  • the evaluation module is based on the And the current control strategy, outputting a cost function
  • the execution module is trained according to the cost function; wherein the evaluation module and the execution module retrained are used for outputting the control strategy at the next moment.
  • control strategy includes: a stored energy correction value.
  • Another embodiment of the present invention provides a computer storage medium having stored therein computer executable instructions for performing any one or more of the methods described above.
  • the embodiment of the invention provides an adaptive optimization control method for a large-scale battery energy storage power station based on adaptive dynamic programming (ADP).
  • ADP adaptive dynamic programming
  • the method and system provided by the embodiments of the present invention comprehensively consider the state of charge of a large-scale battery energy storage power station, the feedback value of new energy power generation fluctuation rate, and the evaluation module and execution module based on a neural network, thereby effectively implementing a large-scale battery energy storage system.
  • the technical solution provided by the embodiment of the invention is based on the neural network to optimize the real-time intelligent optimization of the control algorithm, improve the self-learning and adaptive control ability of the control system, and adaptively dynamically correct the energy storage system in real time to meet the wind power grid connection.
  • SOC state of charge
  • the method can be applied to optimal control of charge and discharge power and battery energy management of large-scale battery energy storage power stations (systems) of different scales.
  • FIG. 1 is a structural diagram of a control system for adaptive dynamic programming of a large-scale energy storage power plant according to an embodiment of the present invention
  • FIG. 2 is a flow chart of a control method for adaptive dynamic programming of a large-scale energy storage power plant according to an embodiment of the present invention.
  • a control system for adaptive dynamic programming of a large-scale energy storage power plant includes:
  • A. Parameter initialization module The adaptive dynamic programming control system selects two layers of evaluation modules and execution modules, each of which is constructed with a three-layer neural network structure.
  • the parameters of the execution module and the evaluation module include the discount factor ⁇ , the network learning rates l a and l c , the weights W a and W c , the maximum number of cycles, and the expected error of the training network.
  • Control target parameters include wind power generation capacity The state of charge SOC limits, sampling time [Delta] t, the observation time T, the fluctuation rate control target storage system capacity W bat, energy storage system Volatility limit target
  • the initial state of the controlled object includes the initial value of the wind power power fluctuation rate, the initial value of the energy storage system state of charge SOC, and the current wind power actual output power.
  • the controlled object here may be the large-scale energy storage power station.
  • B Data acquisition and calculation module. Collect the actual output power of the wind power, charge and discharge power of the energy storage system, calculate the wind power fluctuation rate and the state of charge in real time, and determine in real time whether the state quantity is within the constraint range. When the state quantity is not within the constraint range, adjust the charge and discharge power of the energy storage system.
  • the specificity of the training evaluation module is as follows: Wind power fluctuation rate And the control strategy, that is, the stored energy correction value ⁇ P BESS (t) is sent to the evaluation module as an input, the output J c WPBESS (t) cost function of the evaluation module is calculated, and the objective function E chybrid (t) training evaluation module is constructed, and according to the evaluation The weight of the module is updated, and the weight of the evaluation module neural network is updated; among them, the wind power fluctuation rate And the control strategy, that is, the stored energy correction value ⁇ P BESS (t) should be normalized to [-1, +1], and then sent to the network for calculation.
  • the objective function here is constructed based on the cost function.
  • the training function is re-trained by minimizing the objective function as a training target, and the evaluation module is trained to calculate the wind power volatility based on the next acquisition time. And the control strategy that executes the module output gets the cost function of the next execution module training.
  • the training of the evaluation module updates the weight W c of the evaluation module neural network with the goal of minimizing the objective function E chybrid (t), when the objective function E chybrid (t) decreases to the set error value or the number of iterations reaches the maximum
  • the training is completed.
  • the training execution module trains the execution module according to the output J chybrid (t) of the minimization evaluation module, and updates the weight W a of the execution module neural network according to the weight update formula of the execution module.
  • the execution module is trained by minimizing the output of the evaluation module J chybrid (t) to adjust the control strategy, ie the stored energy correction value ⁇ P BESS (t); updating the execution module by minimizing the objective function E ahybrid (t)
  • the weight of the network, W a is completed when the objective function E ahybrid (t) decreases to a set error value or the number of iterations reaches a maximum.
  • E ahybrid objective function objective function E chybrid (t) trained evaluation module noteworthy and training execution module (t) are different, the E ahybrid (t) is the cost function J chybrid (t) value of n Related, so when the execution module is trained, the execution module is trained for the purpose of minimizing the E ahybrid (t).
  • E output module.
  • the control strategy of each moment is saved and real-time output, and the adjustment and smoothing process is adjusted online in real time to control the charging and discharging power of the energy storage system.
  • the control strategy to control the charge and discharge power of the energy storage system.
  • the present invention comprehensively considers the volatility of new energy power generation output and large-scale battery storage.
  • the real-time power of large-scale energy storage power station can be adjusted in real time through adaptive dynamic programming algorithm, which realizes the comprehensive control of the combined effect of new energy power generation and large-scale energy storage system, and large-scale energy storage.
  • Solving the optimal solution of the charging and discharging power of the energy storage system such as the system capacity.
  • the Adaptive Dynamic Programming (ADP) algorithm does not rely on the mathematical model of the controlled system or process, it has the ability of online self-learning, which can automatically adapt to changes in system parameters and is robust. Therefore, the present invention considers the adaptive dynamic programming algorithm into the adaptive smoothing control, realizes the online adaptive adjustment of the new energy power generation output smoothness, and optimizes the control effect of the large-scale battery energy storage system.
  • a control method for adaptive dynamic programming of a large-scale energy storage power station according to an embodiment of the present invention is provided.
  • the specific steps of the method are as follows:
  • Step 1 Set the structure of the adaptive dynamic programming control system, execute the parameters of the module and the evaluation module, and control the target parameters.
  • the adaptive dynamic programming control system includes a two-layer structure of an evaluation module and an execution module (or a three-layer structure of a model module, an evaluation module, and an execution module), each of which is constructed with a three-layer neural network structure.
  • the parameters of the execution module and the evaluation module include the discount factor ⁇ , the network learning rates l a and l c , the weights W a and W c , the maximum number of cycles, and the expected error of the training network.
  • the initial value of the weight is set to a random value between (-1, +1), and the learning rate and the discount factor are selected and adjusted according to the control effect requirements.
  • the appropriate neural network model (including the type, structure, network parameters and training mode of the network) is selected after collaborative optimization.
  • the action in Figure 2 The network corresponds to the neural network of the execution module of the present application.
  • the evaluation network in Figure 2 is the neural network of the evaluation module.
  • Control target parameters include wind power generation capacity Energy storage system capacity W bat , state of charge SOC limit of energy storage system, sampling time ⁇ t, observation time T, volatility control target Volatility limit target
  • Step 2 initializing parameters and importing an initial state of the controlled object
  • Initializing the parameters and importing the initial state of the controlled object includes the initial value of the wind power fluctuation rate, the initial value of the energy storage state SOC, and the current wind power actual output power.
  • Step 3 Calculate the original wind power fluctuation rate at the current time t Smoothing the original wind power according to the rate of change control strategy; calculating the smoothed wind power fluctuation rate Energy storage system power P BESS (t), state of charge SOC of the energy storage system;
  • Step 4 Initializing a training evaluation module and an execution module
  • the rate of change k(t) of wind power is defined according to the original wind power value P wp (t), the smoothed wind power value P hybrid (t), and the sampling time ⁇ t:
  • P wp (t) is the original wind power value
  • P hybrid (t) is the smoothed wind power value
  • Wind storage power volatility after smoothing rate control strategy defined as:
  • the goal of the rate of change control power volatility method is to control the volatility of the smoothed wind power to be less than a given target value during the observation time:
  • the charging and discharging power of the energy storage system can be calculated according to the above formulas.
  • the energy storage power at time t is:
  • W bat is the capacity of the energy storage system.
  • Step 5 Determine the wind power fluctuation rate Whether it is in the constraint If it is not within the constraint range, proceed to the next ADP network training to seek the optimal stored energy correction value; if it is within the constraint range, the energy storage system will not operate and the stored energy will not be corrected;
  • Step 6 The state of the controlled object, that is, the wind power fluctuation rate
  • the energy storage power P BESS (t) acts as an input to the execution network, trains the execution network, updates the weight of the execution network, and outputs the control strategy, that is, the correction value of the stored energy ⁇ P BESS (t);
  • the training execution module trains the execution module according to the output J chybrid (t) of the minimization evaluation module, and updates the weight W a of the execution module neural network according to the weight update formula of the execution module.
  • the execution module is trained by minimizing the output of the evaluation module J chybrid (t) to adjust the control strategy, ie the stored energy correction value ⁇ P BESS (t); updating the execution module by minimizing the objective function E ahybrid (t) network weight value W a, when the objective function E ahybrid (t) or an error value is reduced to a set number of iterations reaches the maximum, the training is completed.
  • Step 7 The controlled object state and the control strategy are used as input of the evaluation network, the evaluation network is trained, the weight of the evaluation network is updated, and the output is a cost function;
  • Wind power volatility And the control strategy that is, the stored energy correction value ⁇ P BESS (t) is sent to the evaluation module as an input, the output J c WPBESS (t) cost function of the evaluation module is calculated, and the objective function E chybrid (t) training evaluation module is constructed, and according to the evaluation The weight of the module is updated, and the weight of the evaluation module neural network is updated; among them, the wind power fluctuation rate And the control strategy, that is, the stored energy correction value ⁇ P BESS (t) should be normalized to [-1, +1], and then sent to the network for calculation.
  • the training of the evaluation module updates the weight W c of the evaluation module neural network with the goal of minimizing the objective function E chybrid (t), when the objective function E chybrid (t) decreases to the set error value or the number of iterations reaches the maximum
  • the training is completed.
  • Step 9 Loop the above steps until the control process ends, and output the control strategy at each moment, and the smoothed wind power fluctuation rate Energy storage power And energy storage system SOC.
  • the embodiment of the invention further provides another control method for adaptive dynamic programming of a large-scale energy storage power plant, the method comprising the following steps:
  • the current execution module is based on the An output control strategy, wherein the control strategy is used to control charging and discharging power of an energy storage system of the large-scale energy storage power station;
  • the evaluation module is based on the And the current control strategy, outputting a cost function
  • the execution module is trained according to the cost function; wherein the evaluation module and the execution module retrained are used for outputting the control strategy at the next moment.
  • the P BESS (t) may include charging power and discharging power
  • the charging power is between the maximum allowable charging power and the minimum power
  • the discharging power is at the maximum allowable discharging power and Between the minimum amplification powers, it is considered to be within the constraint range.
  • the minimum charging power and the amplified discharging power are both zero.
  • the SOC is less than the maximum state of charge allowed during the operation of the spring energy system, and is greater than the minimum state of charge allowed during operation, and the SOC is within the constraint range.
  • the above is considered Located within the constraint range; only when all three parameters are within the corresponding constraint range, it is considered that there is no need to adjust at present, and no control strategy is required, otherwise one or more of these parameters are input to the current execution module.
  • control the charging and discharging power of the energy storage system and re-train the execution module and evaluation module to facilitate more accurate control of the energy storage system of the large-scale energy storage power station.
  • the control strategy includes: a stored energy correction value.
  • the energy storage system of the large-scale energy storage power station adjusts its own charging and discharging power according to the stored energy correction value.
  • Another embodiment of the present invention provides a computer storage medium having stored therein computer executable instructions for performing any one or more of the methods described above.
  • the computer storage medium described in this embodiment may be various types of storage media, and may be a non-transitory storage medium.
  • a control system for adaptive dynamic programming of a large-scale energy storage power plant wherein the system includes a parameter initialization module, a data acquisition and calculation module, an execution module, an evaluation module, and an output module;
  • the parameter initialization module is configured to set a structure and a control target parameter of the adaptive dynamic programming control system, and transmit the target parameter to the data collection and calculation module;
  • the data collection and calculation module is configured to calculate a wind storage power fluctuation rate according to the target parameter, and transmit the wind storage power fluctuation rate to the execution module and the evaluation module;
  • the execution module is configured to obtain a stored energy correction value according to the wind storage power fluctuation rate, and transmit the stored energy correction value to the data acquisition and calculation module, the evaluation module, and the output mode;
  • the evaluation module is configured to derive a cost function according to the wind power fluctuation rate and transmit the cost function to the execution module;
  • the output module is configured to output the control strategy at each moment, the smoothed wind storage power fluctuation rate, the energy storage power, and the energy storage system state of charge SOC.

Landscapes

  • Engineering & Computer Science (AREA)
  • Power Engineering (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Theoretical Computer Science (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Physics (AREA)
  • Mechanical Engineering (AREA)
  • Combustion & Propulsion (AREA)
  • Chemical & Material Sciences (AREA)
  • Sustainable Energy (AREA)
  • Sustainable Development (AREA)
  • Biomedical Technology (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Automation & Control Theory (AREA)
  • Medical Informatics (AREA)
  • Fuzzy Systems (AREA)
  • Supply And Distribution Of Alternating Current (AREA)
  • Control Of Eletrric Generators (AREA)
  • Wind Motors (AREA)

Abstract

一种大规模储能电站自适应动态规划的控制方法、***和存储介质，所述方法包括：设置自适应动态规划控制***的结构和控制目标参数；初始化参数并导入被控对象的初始状态；计算当前时刻t的原始风电功率波动率，根据变化率控制策略对原始风电功率进行平滑；计算平滑后的风储功率波动率，储能***功率，储能***的荷电状态SOC；初始化训练评价模块和执行模块；计算每个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC，并进行保存；输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。所述***包括参数初始化模块、数据采集和计算模块、执行模块、评价模块及输出模块等。

Description

储能电站自适应动态规划的控制方法、***和存储介质
本申请基于申请号为201610278732.6、申请日为2016年04月28日的中国专利申请提出，并要求该中国专利申请的优先权，该中国专利申请的全部内容在此引入本申请作为参考。
技术领域
本发明属于智能电网、能源互联网和储能技术领域，具体涉及一种大规模储能电站自适应动态规划的控制方法、***和存储介质。
背景技术
随着风电、光伏发电技术的不断发展，风电和光伏发电等新能源发电的大规模并网，其出力的波动问题日益严重。风电、光伏发电等新能源发电大规模接入电网后，其波动性和间歇性会对电网运行的安全性、稳定性以及电能质量等造成不利影响。因此，目前在实际运用中，风电和光伏发电的并网受到了很大限制，不利于风电、光伏发电等新能源发电的发展，控制新能源发电出力波动对电网运行的安全、稳定、经济运行有着重要意义。储能***凭借充放电能力可有效抑制新能源发电输出波动对电网的不利影响，进而降低新能源发电***带来的波动性，提高电网接纳新能源发电的能力。
按照存储形式的不同，储能可分为物理储能、电化学储能和电磁储能。其中电池储能目前处于快速发展阶段，储能电站规模达到了兆瓦级～数十兆瓦级。因此可以通过配备一定容量的大规模电池储能***，根据新能源发电出力的情况，采用电池储能***充放电的优化控制，平滑新能源发电功率。并结合新能源发电出力波动率和储能***的荷电状态等指标进行整体优化，以满足电网对风电、光伏发电等新能源发电并网的要求。
目前，我国已建成多个千万千瓦级新能源发电基地，在新能源发电富集区域电网中，对电池储能的容量要求通常达数十MW以上，甚至达百兆瓦以上。百兆瓦级电池储能电站参与新能源发电集群控制及***调度运行，对破解新能源发电的送出和消纳瓶颈有重要意义。大规模储能技术是我国可再生能源发电利用的关键支撑技术。针对大规模可再生能源发电的接入，一方面通过储能技术与可再生能源发电的联合，减少其随机性并提高其可调性；另一方面通过电网级的储能应用增强电网对可再生能源发电的适应性。目前，储能作为电网的可调度资源，具有很大的应用价值和应用空间。
在电网级应用中，需要储能进行秒至小时级的多时间尺度功率支撑。在储能与新能源发电联合并网应用时，百兆瓦电池储能电站整体需满足新能源发电从秒到分钟级不同时间尺度的响应需求。因此，如何基于大规模新能源发电的出力波动以及储能在电网级应用等实际需求，实现百兆瓦级电池储能电站整体出力多目标协调优化控制是亟待破解的技术难题。
在大规模电池储能电站平滑大规模新能源发电出力波动时，采用传统的一阶低通滤波或变时间常数(T)的出力滤波方法，由于方法本身自带的时滞而导致有时控制灵敏度不佳。一般的移动平均滤波算法输出的储能出力目标值也在很大程度上受限于风电、光伏发电等新能源发电功率的输入，这些现有方法遇到新能源发电出力出现骤变的情况，滤波性能下降，且影响后续滤波效果。另一方面，在提高新能源发电友好性的储能电站出力控制时，传统控制方法在储能电站整体出力的自适应控制方面，其基于自学习的智能优化控制能力有待进一步提高。
发明内容
本发明实施例提供一种大规模储能电站自适应动态规划的控制方法、***和存储介质，期望降低风电功率并网给电网带来的冲击，同时对储能***的工作能力和寿命进行优化保护，以提升储能***的技术性与经济性。
本发明实施例采取如下技术方案：
一种大规模储能电站自适应动态规划的控制方法，所述方法包括如下步骤：
(1)设置自适应动态规划控制***的结构和控制目标参数；
(2)初始化参数并导入被控对象的初始状态；
(3)计算当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000001
根据变化率控制方法对原始风电功率进行平滑；计算平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000002
储能***功率PBESS(t)，储能***的荷电状态SOC；
(4)初始化训练评价模块和执行模块；
(5)计算每个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC，并进行保存；
(6)输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。
可选地，所述步骤(1)中，所述自适应动态规划控制***包括评价模块和执行模块的两层结构，每个模块均用三层神经网络结构来构建；所述控制目标参数包括风电发电容量
Figure PCTCN2017082564-appb-000003
储能***容量Wbat，储能***的荷电状态SOC限制范围，采样时间Δt，观测时间T，波动率控制目标
Figure PCTCN2017082564-appb-000004
和波动率限制目标
Figure PCTCN2017082564-appb-000005
可选地，所述步骤(2)中，所述初始化参数包括风电功率波动率的初
始值，储能***荷电状态SOC的初始值，当前时刻风电实际出力功率。
可选地，所述步骤(3)包括如下步骤：
步骤3-1、所述的当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000006
的计算公式如下：
Figure PCTCN2017082564-appb-000007
Figure PCTCN2017082564-appb-000008
Figure PCTCN2017082564-appb-000009
T＝nΔt           (4)
式中：
Figure PCTCN2017082564-appb-000010
是风电容量，即额定功率，
Figure PCTCN2017082564-appb-000011
和
Figure PCTCN2017082564-appb-000012
是观测时间内的风电功率最大值和最小值，Δt是采样时间，T为观测时间，Pwp(t)为风电功率的原始值,fwp计算风电功率波动率的原始函数,n为在观测时间内的采样点数；
步骤3-2、根据变化率控制方法对所述风电功率的原始值进行平滑；
所述变化率控制方法的计算方法如下：
根据风电功率原始值Pwp(t)、平滑后的风电功率值Phybrid(t)、采样时间Δt定义了风电功率的变化率k(t)：
Figure PCTCN2017082564-appb-000013
式中：Pwp(t)是风电功率原始值；Phybrid(t)是平滑后的风储功率值；
为了使风储功率波动率控制在要求范围内，制定如下控制策略：
如果
Figure PCTCN2017082564-appb-000014
Phybrid(t)＝Pwp(t-Δt)        (6)
如果
Figure PCTCN2017082564-appb-000015
Figure PCTCN2017082564-appb-000016
如果
Figure PCTCN2017082564-appb-000017
Figure PCTCN2017082564-appb-000018
式中：
Figure PCTCN2017082564-appb-000019
是风电出力的上升阶段的变化率的限制值，
Figure PCTCN2017082564-appb-000020
是风电出力的下降阶段的变化率的限制值；
其中
Figure PCTCN2017082564-appb-000021
Figure PCTCN2017082564-appb-000022
式中，
Figure PCTCN2017082564-appb-000023
为波动率控制目标；
步骤3-3、经过所述变化率控制策略平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000024
的公式为：
Figure PCTCN2017082564-appb-000025
Figure PCTCN2017082564-appb-000026
Figure PCTCN2017082564-appb-000027
T＝nΔt         (14)
式中：
Figure PCTCN2017082564-appb-000028
和
Figure PCTCN2017082564-appb-000029
是观测时间T内的风储功率最大值和最小值；
变化率控制功率波动率方法的目标是控制平滑后的风储功率波动率在观测时间内小于给定目标值：
Figure PCTCN2017082564-appb-000030
式中：
Figure PCTCN2017082564-appb-000031
是给定观测时间T内风储功率波动率的控制目标；
步骤3-4、所述储能***功率的计算公式为：
PBESS(t)＝Phybrid(t)-Pwp(t)       (16)
所述储能***的荷电状态SOC的计算公式如下：
如果PBESS(t)＞0，储能***放电，SOC减小，
Figure PCTCN2017082564-appb-000032
如果PBESS(t)＜0，储能***充电，SOC增大，
Figure PCTCN2017082564-appb-000033
式中：Wbat是储能***容量。
可选地，所述步骤(4)中，所述初始化训练评价模块和执行模块包括：初始化设置折扣因子α、执行模块学习率la和评价模块学习率lc、执行模块权值Wa和评价模块权值Wc，最大循环次数，以及训练模块的期望误差，其中权值的初始值设置为(-1,+1)间的随机值，学习率和折扣因子依 据控制效果要求进行选取和调整。
可选地，所述步骤(5)包括如下步骤：
步骤5-1、判断风储功率波动率
Figure PCTCN2017082564-appb-000034
是否在约束条件
Figure PCTCN2017082564-appb-000035
内，若是则储能***不动作，不对储能功率进行修正；否则进行下一步的自适应动态规划模块训练，寻求最优储能功率修正值；
步骤5-2、将被控对象状态和所述控制策略作为评价模块的输入，训练评价模块，更新评价模块的权值，输出代价函数；
步骤5-3、将被控对象状态包括风储功率波动率
Figure PCTCN2017082564-appb-000036
和储能功率PBESS(t)作为执行模块的输入，训练执行模块，更新执行模块的权值，输出为控制策略，即储能功率的修正值ΔPBESS(t)；
步骤5-4、保存此时刻的控制策略，并计算下一时刻被控对象状态，t＝t+1，重复步骤5-1到步骤5-3，直到控制过程结束。
可选地，所述步骤5-1中，所述判断风储功率波动率
Figure PCTCN2017082564-appb-000037
的约束如下：
如果
Figure PCTCN2017082564-appb-000038
则储能***过度出力，需要进行反向修正：
Figure PCTCN2017082564-appb-000039
式中，ΔPBESS(t)为储能功率修正值
如果
Figure PCTCN2017082564-appb-000040
储能***出力适宜，无需修正：
Figure PCTCN2017082564-appb-000041
式中：
Figure PCTCN2017082564-appb-000042
是在变化率控制策略的基础上，对储能功率进行自适应动态规划调节后的储能***的功率。
可选地，所述步骤5-2包括如下步骤：
步骤5-2-1、将所述被控对象状态即风储功率波动率
Figure PCTCN2017082564-appb-000043
和所述控制策略即储能功率修正值ΔPBESS(t)进行归一化处理为[-1，+1]之间；
步骤5-2-2、将所述被控对象状态即风储功率波动率
Figure PCTCN2017082564-appb-000044
和所述控制策略即储能功率修正值ΔPBESS(t)送入所述评价模块作为输入，计算所述评价模 块的输出Jc WPBESS(t)代价函数，构造目标函数Echybrid(t)训练所述评价模块，并根据评价模块的权值更新式，更新评价模块神经网络的权值，计算公式如下：
Figure PCTCN2017082564-appb-000045
Figure PCTCN2017082564-appb-000046
Figure PCTCN2017082564-appb-000047
式中：代价函数Jchybrid(t)为评价模块的输出，效用函数U(t)是关于
Figure PCTCN2017082564-appb-000048
的函数，根据控制目标进行定义，βc是折扣因子；
步骤5-2-3、所述评价模块的训练采用梯度下降法或粒子群优化算法以使目标函数Echybrid(t)最小化为目标来更新评价模块神经网络的权值Wc，当目标函数Echybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
可选地，所述步骤5-3包括如下步骤：
步骤5-3-1、将被控对象状态包括风储功率波动率
Figure PCTCN2017082564-appb-000049
和储能功率PBESS(t)作为执行模块的输入，训练执行模块；
步骤5-3-2、通过最小化评价模块输出Jchybrid(t)来调整控制策略即储能功率修正值ΔPBESS(t)，公式如下：
Figure PCTCN2017082564-appb-000050
Figure PCTCN2017082564-appb-000051
式中：控制策略ΔPBESS(t)是执行模块的输出，用来调整储能功率在合理范围内变化，以减小储能***SOC波动范围，u表示控制策略ΔPBESS(t)是关于
Figure PCTCN2017082564-appb-000052
的函数；
步骤5-3-3、所述执行模块的训练采用梯度下降法或粒子群优化算法以最小化目标函数Eahybrid(t)为目标来更新执行模块神经网络的权值Wa，当目标函数Eahybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
可选地，一种大规模储能电站自适应动态规划的控制***，所述*** 包括参数初始化模块、数据采集和计算模块、执行模块、评价模块及输出模块；
所述参数初始化模块，配置为设置自适应动态规划控制***的结构和控制目标参数，并将所述目标参数传送到所述数据采集和计算模块；
所述数据采集和计算模块，配置为根据所述目标参数计算风储功率波动率，并将所述风储功率波动率传送到所述执行模块和评价模块；
所述执行模块，配置为根据所述风储功率波动率得到储能功率修正值，并将所述储能功率修正值传送到所述数据采集和计算模块、评价模块及输出模；
所述评价模块，配置为根据所述风储功率波动率得出代价函数，并将其传送到执行模块；
所述输出模块，配置为输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。
本发明实施例还提供另一种一种大规模储能电站自适应动态规划的控制方法，所述方法包括如下步骤：
计算当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000053
根据变化率控制策略对原始风电功率进行平滑，计算平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000054
并获得所储能***功率PBESS(t)，储能***的荷电状态SOC；
判断所述
Figure PCTCN2017082564-appb-000055
所述PBESS(t)及所述SOC是否在控制目标参数对应的约束范围内；
当所述
Figure PCTCN2017082564-appb-000056
所述PBESS(t)及所述SOC不在所述约束范围内时，将所述
Figure PCTCN2017082564-appb-000057
输入到当前的执行模块；
当前的所述执行模块基于所述
Figure PCTCN2017082564-appb-000058
输出控制策略，其中，所述控制策 略用于控制所述大规模储能电站的储能***的充放电功率；
将所述
Figure PCTCN2017082564-appb-000059
和当前控制策略输入到当前评价模块；
所述评价模块基于所述
Figure PCTCN2017082564-appb-000060
及所述当前控制策略时，输出代价函数；
基于所述
Figure PCTCN2017082564-appb-000061
及所述代价函数，构建训练所述评价模块的目标函数；
以最小化所述目标函数为训练目的，依据所述目标函数训练所述评价模块；
以最小化所述代价函数为目的，依据所述代价函数训练所述执行模块；其中，重新训练的所述评价模块和执行模块，用于下一时刻的所述控制策略的输出。
基于上述方案，所述控制策略包括：储能功率修正值。
本发明实施例还提供另一种计算机存储介质，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令用于执行上述任意一个或多个方法。
本发明实施例提出了一种基于自适应动态规划(ADP)的大型电池储能电站自适应优化控制方法。本发明实施例提供给的方法及***综合考虑大规模电池储能电站荷电状态、新能源发电波动率反馈值以及基于神经网络的评价模块和执行模块等，有效实现了大规模电池储能***整体充放电功率的自适应优化控制。本发明实施例提供的技术方案基于神经网络对控制算法进行实时智能优化，提高了控制***自学习和自适应控制能力，实时对储能***出力进行自适应动态修正，使其满足风电并网的要求，同时，又控制储能电池荷电状态(SOC)保持在适宜的范围内，实现储能电池***的合理充放电的功能，满足大规模储能***实时充放电功率的优化控制目的。该方法可适用于不同规模等级的大型电池储能电站(***)的充放电功率的优化控制与电池能量管理。
附图说明
图1是本发明实施例提供的一种大规模储能电站自适应动态规划的控制***的结构图，
图2是本发明实施例提供的一种大规模储能电站自适应动态规划的控制方法的流程图。
具体实施方式
下面结合附图对本发明实施例作进一步详细说明，应当理解，以下所说明的优选实施例仅用于说明和解释本发明，并不用于限定本发明。
如图1所示，为本发明实施例提供的一种大规模储能电站自适应动态规划的控制***，该***包括：
A、参数初始化模块。自适应动态规划控制***选择评价模块和执行模块两层结构，每个模块均用三层神经网络结构来构建。执行模块和评价模块的参数包括折扣因子α、网络学习率la和lc、权值Wa和Wc，最大循环次数，以及训练网络的期望误差。
控制目标参数包括风电发电容量
Figure PCTCN2017082564-appb-000062
储能***容量Wbat，储能***的荷电状态SOC限制范围，采样时间Δt，观测时间T，波动率控制目标
Figure PCTCN2017082564-appb-000063
波动率限制目标
Figure PCTCN2017082564-appb-000064
被控对象的初始状态包括风电功率波动率的初始值，储能***荷电状态SOC的初始值，当前时刻风电实际出力功率。这里的被控对象可为所述大规模储能电站。
B、数据采集和计算模块。采集风电实际出力功率，储能***充放电功率，实时计算风电波动率和荷电状态，并实时判断状态量是否在约束范围内。当状态量不在约束范围内时，调整储能***充放电功率。
C、评价模块。训练评价模块的具体如下：将风电波动率
Figure PCTCN2017082564-appb-000065
和控制策 略即储能功率修正值ΔPBESS(t)送入评价模块作为输入，计算评价模块的输出Jc WPBESS(t)代价函数，构造目标函数Echybrid(t)训练评价模块，并根据评价模块的权值更新式，更新评价模块神经网络的权值；其中，风电波动率
Figure PCTCN2017082564-appb-000066
和控制策略即储能功率修正值ΔPBESS(t)都应经归一化处理为[-1，+1]后，再送入网络进行计算。这里的目标函数是基于所述代价函数构建的。在构建出所述目标函数Echybrid(t)之后，以最小化所述目标函数为训练目标，重新训练所述评价模块，训练好评价模块用于基于下一个采集时刻采集的风电波动率
Figure PCTCN2017082564-appb-000067
和执行模块输出的控制策略获得下一次执行模块训练的代价函数。
评价模块的训练以最小化目标函数Echybrid(t)为目标来更新评价模块神经网络的权值Wc，当目标函数Echybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
D、执行模块。训练执行模块的具体步骤如下：训练执行模块是依据最小化评价模块的输出Jchybrid(t)来训练执行模块，并根据执行模块的权值更新式，更新执行模块神经网络的权值Wa。执行模块的训练是通过最小化评价模块的输出Jchybrid(t)来调整控制策略即储能功率修正值ΔPBESS(t)；通过最小化目标函数Eahybrid(t)为目标来更新执行模块神经网络的权值Wa，当目标函数Eahybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
值得注意的训练评价模块的目标函数Echybrid(t)与训练执行模块的目标函数Eahybrid(t)是不同的，所述Eahybrid(t)是与代价函数Jchybrid(t)的取值正相关的，故即在训练所述执行模块时，是以最小化所述Eahybrid(t)为目的训练所述执行模块的。
E、输出模块。保存并实时输出各时刻的控制策略，实时在线调整调整平滑过程，控制储能***充放电功率。这里是利用控制策略控制储能***的充放电功率。
综上所述，本发明在综合考虑新能源发电出力波动率和大规模电池储 能***荷电状态的条件下，通过自适应动态规划算法实时调节大规模储能电站实时功率，实现了综合考虑新能源发电与大规模储能***联合并网运行的控制效果、大规模储能***容量等的储能***充放电功率的最优解的求解。由于自适应动态规划(ADP:Adaptive Dynamic Programming)算法不依赖被控***或者过程精确的数学模型，具有在线自学***滑控制中，实时了新能源发电出力平滑的在线自适应调节，优化大规模电池储能***的控制效果。
如图2所示，为本发明实施例提供的一种大规模储能电站自适应动态规划的控制方法，该方法具体步骤如下：
步骤1、设置自适应动态规划控制***的结构，执行模块和评价模块的参数,以及控制目标参数。
自适应动态规划控制***包括评价模块和执行模块的两层结构，(或者模型模块、评价模块和执行模块的三层结构)，每个模块均用三层神经网络结构来构建。执行模块和评价模块的参数包括折扣因子α、网络学习率la和lc、权值Wa和Wc，最大循环次数，以及训练网络的期望误差。其中权值的初始值设置为(-1,+1)间的随机值，学习率和折扣因子依据控制效果要求进行选取和调整。综合考虑被控对象的具体情况和神经网络的收敛速度、计算结果的精确度，协同优化后选取合适的神经网络模型(包括网络的类型、结构、网络参数和训练模式)，图2中的动作网络对应于本申请的执行模块的神经网络。图2中的评价网络为评价模块的神经网络。
控制目标参数包括风电发电容量
Figure PCTCN2017082564-appb-000068
储能***容量Wbat，储能***的荷电状态SOC限制范围，采样时间Δt，观测时间T，波动率控制目标
Figure PCTCN2017082564-appb-000069
波动率限制目标
Figure PCTCN2017082564-appb-000070
步骤2、初始化参数并导入被控对象的初始状态；
初始化参数并导入被控对象的初始状态包括风电功率波动率的初始值，储能***荷电状态SOC的初始值，当前时刻风电实际出力功率。
步骤3、计算当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000071
根据变化率控制策略对原始风电功率进行平滑；计算平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000072
储能***功率PBESS(t)，储能***的荷电状态SOC；
步骤4、初始化训练评价模块和执行模块；
计算当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000073
利用变化率控制策略对原始功率波动率进行平滑，计算方法如下：
根据风电功率原始值Pwp(t)、平滑后的风储功率值Phybrid(t)、采样时间Δt定义了风电功率的变化率k(t)：
Figure PCTCN2017082564-appb-000074
式中：Pwp(t)是风电功率原始值；Phybrid(t)是平滑后的风储功率值。
为了使风储功率波动率控制在要求范围内，制定如下控制策略：
如果
Figure PCTCN2017082564-appb-000075
Phybrid(t)＝Pwp(t-Δt)        (6)
如果
Figure PCTCN2017082564-appb-000076
Figure PCTCN2017082564-appb-000077
如果
Figure PCTCN2017082564-appb-000078
Figure PCTCN2017082564-appb-000079
其中：
Figure PCTCN2017082564-appb-000080
是风电出力的上升阶段的变化率的限制值；
Figure PCTCN2017082564-appb-000081
是风电出力的下降阶段的变化率的限制值，二者定义为：
Figure PCTCN2017082564-appb-000082
Figure PCTCN2017082564-appb-000083
经过变化率控制策略平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000084
定义为：
Figure PCTCN2017082564-appb-000085
Figure PCTCN2017082564-appb-000086
Figure PCTCN2017082564-appb-000087
T＝nΔt       (14)
上四式中：
Figure PCTCN2017082564-appb-000088
和
Figure PCTCN2017082564-appb-000089
是观测时间T内的风储功率最大值和最小值。
变化率控制功率波动率的方法的目标是控制平滑后的风储功率的波动率在观测时间内小于给定目标值：
Figure PCTCN2017082564-appb-000090
式中：
Figure PCTCN2017082564-appb-000091
是给定观测时间T内风储功率波动率的控制目标。
储能***充放电功率可以根据上面几式计算得到，在t时刻的储能功率为：
PBESS(t)＝Phybrid(t)-Pwp(t)       (16)
储能***的荷电状态SOC的计算公式如下：
如果PBESS(t)＞0，储能***放电，SOC减小。
Figure PCTCN2017082564-appb-000092
如果PBESS(t)＜0，储能***充电，SOC增大。
Figure PCTCN2017082564-appb-000093
其中：Wbat是储能***容量。
步骤5、判断风储功率波动率
Figure PCTCN2017082564-appb-000094
是否在约束条件
Figure PCTCN2017082564-appb-000095
内，若不在约束范围内，则进行下一步的ADP网络训练，寻求最优储能功率修正值；若在约束范围内，则储能***不动作，不对储能功率进行修正；
确定ADP是否需要调节储能功率的具体操作为：基于储能***的储能功率PBESS(t)和变化率控制后的风储功率波动率
Figure PCTCN2017082564-appb-000096
引入一个新的限制风储功率波动率的值，定义该值为
Figure PCTCN2017082564-appb-000097
结合该限制值和波动率控制目标
Figure PCTCN2017082564-appb-000098
制 定以下的控制策略：
如果
Figure PCTCN2017082564-appb-000099
储能***过度出力，需要进行反向修正。
Figure PCTCN2017082564-appb-000100
如果
Figure PCTCN2017082564-appb-000101
储能***出力适宜，无需修正。
Figure PCTCN2017082564-appb-000102
式中：
Figure PCTCN2017082564-appb-000103
是在变化率控制策略的基础上，对储能功率进行ADP调节后的储能***的功率。
步骤6、将被控对象状态，即风储功率波动率
Figure PCTCN2017082564-appb-000104
储能功率PBESS(t)作为执行网络的输入，训练执行网络，更新执行网络的权值，输出为控制策略，即储能功率的修正值ΔPBESS(t)；
训练执行网络的具体步骤如下：
训练执行模块是依据最小化评价模块的输出Jchybrid(t)来训练执行模块，并根据执行模块的权值更新式，更新执行模块神经网络的权值Wa。执行模块的训练是通过最小化评价模块的输出Jchybrid(t)来调整控制策略即储能功率修正值ΔPBESS(t)；通过最小化目标函数Eahybrid(t)为目标来更新执行模块神经网络的权值Wa，当目标函数Eahybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
步骤7、将被控对象状态和控制策略作为评价网络的输入，训练评价网络，更新评价网络的权值，输出为代价函数；
训练评价网络的具体步骤如下：
将风电波动率
Figure PCTCN2017082564-appb-000105
和控制策略即储能功率修正值ΔPBESS(t)送入评价模块作为输入，计算评价模块的输出Jc WPBESS(t)代价函数，构造目标函数Echybrid(t)训练评价模块，并根据评价模块的权值更新式，更新评价模块神经网络的权值；其中，风电波动率
Figure PCTCN2017082564-appb-000106
和控制策略即储能功率修正值ΔPBESS(t)都应经归一化处理为[-1，+1]后，再送入网络进行计算。
评价模块的训练以最小化目标函数Echybrid(t)为目标来更新评价模块神经网络的权值Wc，当目标函数Echybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
步骤8、保存此时刻的控制策略，并计算下一时刻被控对象的状态，t＝t+1，重复步骤5到7；
步骤9、循环上述步骤，直至控制过程结束，并输出各时刻的控制策略，平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000107
储能功率
Figure PCTCN2017082564-appb-000108
和储能***SOC。
本发明实施例还提供另一种一种大规模储能电站自适应动态规划的控制方法，所述方法包括如下步骤：
计算当前时刻t的原始风电功率波动率
Figure PCTCN2017082564-appb-000109
根据变化率控制策略对原始风电功率进行平滑，计算平滑后的风储功率波动率
Figure PCTCN2017082564-appb-000110
并获得所储能***功率PBESS(t)，储能***的荷电状态SOC；
判断所述
Figure PCTCN2017082564-appb-000111
所述PBESS(t)及所述SOC是否在控制目标参数对应的约束范围内；
当所述
Figure PCTCN2017082564-appb-000112
所述PBESS(t)及所述SOC不在所述约束范围内时，将所述
Figure PCTCN2017082564-appb-000113
输入到当前的执行模块；
当前的所述执行模块基于所述
Figure PCTCN2017082564-appb-000114
输出控制策略，其中，所述控制策略用于控制所述大规模储能电站的储能***的充放电功率；
将所述
Figure PCTCN2017082564-appb-000115
和当前控制策略输入到当前评价模块；
所述评价模块基于所述
Figure PCTCN2017082564-appb-000116
及所述当前控制策略时，输出代价函数；
基于所述
Figure PCTCN2017082564-appb-000117
及所述代价函数，构建训练所述评价模块的目标函数；
以最小化所述目标函数为训练目的，依据所述目标函数训练所述评价 模块；
以最小化所述代价函数为目的，依据所述代价函数训练所述执行模块；其中，重新训练的所述评价模块和执行模块，用于下一时刻的所述控制策略的输出。
在本实施例中当所述PBESS(t)可包括充电功率和放电功率，当所述充电功率位于允许的最大充电功率和最小功率之间，所述放电功率位于允许的最大的放电功率和最小的放大功率之间，则认为位于所述约束范围内。通常最小充电功率和放大放电功率均为0。
所述SOC小于春能***工作时允许的最大荷电状态，大于工作时允许的最小荷电状态，则SOC位于约束范围内。
所述
Figure PCTCN2017082564-appb-000118
小于目标值，则认为所述
Figure PCTCN2017082564-appb-000119
位于约束范围内；只有当三个参数都位于对应的约束范围内时，则认为是当前不需要调整，不需要输出控制策略，否则将这些参数中的一个或多个输入到当前的执行模块，以输出当前的控制策略，控制储能***的充放电功率，并重新训练执行模块及评价模块，方便后续更加精确的控制大规模储能电站的储能***。
所述控制策略包括：储能功率修正值。所述大规模储能电站的储能***，根据所述储能功率修正值，调整自身的充放电功率。
本发明实施例还提供另一种计算机存储介质，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令用于执行上述任意一个或多个方法。
本实施例所述的计算机存储介质可为各种类型的存储介质，可选为非瞬间存储介质。
如图1所示，种大规模储能电站自适应动态规划的控制***，其中，所述***包括参数初始化模块、数据采集和计算模块、执行模块、评价模块及输出模块；
所述参数初始化模块，配置为设置自适应动态规划控制***的结构和控制目标参数，并将所述目标参数传送到所述数据采集和计算模块；
所述数据采集和计算模块，用于根据所述目标参数计算风储功率波动率，并将所述风储功率波动率传送到所述执行模块和评价模块；
所述执行模块，配置为根据所述风储功率波动率得到储能功率修正值，并将所述储能功率修正值传送到所述数据采集和计算模块、评价模块及输出模；
所述评价模块，配置为根据所述风储功率波动率得出代价函数，并将其传送到执行模块；
所述输出模块，配置为输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。
最后应当说明的是：以上实施例仅用以说明本发明的技术方案而非对其限制，但凡按照本发明原理所作的修改，都应当理解为落入本发明的保护范围。

Claims (13)

  1. 一种大规模储能电站自适应动态规划的控制方法，所述方法包括如下步骤：
    (1)设置自适应动态规划控制***的结构和控制目标参数；
    (2)初始化参数并导入被控对象的初始状态；
    (3)计算当前时刻t的原始风电功率波动率
    Figure PCTCN2017082564-appb-100001
    根据变化率控制方法对原始风电功率进行平滑；计算平滑后的风储功率波动率
    Figure PCTCN2017082564-appb-100002
    储能***功率PBESS(t)，储能***的荷电状态SOC；
    (4)初始化训练评价模块和执行模块；
    (5)计算每个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC，并进行保存；
    (6)输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。
  2. 根据权利要求1所述控制方法，其中，所述步骤(1)中，所述自适应动态规划控制***包括评价模块和执行模块的两层结构，每个模块均用三层神经网络结构来构建；所述控制目标参数包括风电发电容量
    Figure PCTCN2017082564-appb-100003
    储能***容量Wbat、储能***的荷电状态SOC限制范围、采样时间Δt、观测时间T、波动率控制目标和波动率限制目标
    Figure PCTCN2017082564-appb-100005
  3. 根据权利要求1所述控制方法，其中，所述步骤(2)中，所述初始化参数包括风电功率波动率的初始值，储能***荷电状态SOC的初始值，当前时刻风电实际出力功率。
  4. 根据权利要求1所述控制方法，其中，所述步骤(3)包括如下步骤：
    步骤3-1、所述的当前时刻t的原始风电功率波动率
    Figure PCTCN2017082564-appb-100006
    的计算公式如下：
    Figure PCTCN2017082564-appb-100007
    Figure PCTCN2017082564-appb-100008
    Figure PCTCN2017082564-appb-100009
    T＝nΔt    (4)
    式中：
    Figure PCTCN2017082564-appb-100010
    是风电容量，即额定功率，
    Figure PCTCN2017082564-appb-100011
    和
    Figure PCTCN2017082564-appb-100012
    是观测时间内的风电功率最大值和最小值，Δt是采样时间，T为观测时间，Pwp(t)为风电功率的原始值,fwp计算风电功率波动率的原始函数,n为在观测时间内的采样点数；
    步骤3-2、根据变化率控制方法对所述风电功率的原始值进行平滑；
    所述变化率控制方法的计算方法如下：
    根据风电功率原始值Pwp(t)、平滑后的风电功率值Phybrid(t)、采样时间Δt定义了风电功率的变化率k(t)：
    Figure PCTCN2017082564-appb-100013
    式中：Pwp(t)是风电功率原始值；Phybrid(t)是平滑后的风储功率值；
    为了使风储功率波动率控制在要求范围内，制定如下控制策略：
    如果
    Figure PCTCN2017082564-appb-100014
    Phybrid(t)＝Pwp(t-Δt)    (6)
    如果
    Figure PCTCN2017082564-appb-100015
    Figure PCTCN2017082564-appb-100016
    如果
    Figure PCTCN2017082564-appb-100017
    Figure PCTCN2017082564-appb-100018
    式中：
    Figure PCTCN2017082564-appb-100019
    是风电出力的上升阶段的变化率的限制值，
    Figure PCTCN2017082564-appb-100020
    是风电出力的下降阶段的变化率的限制值；
    其中
    Figure PCTCN2017082564-appb-100021
    Figure PCTCN2017082564-appb-100022
    式中，
    Figure PCTCN2017082564-appb-100023
    为波动率控制目标；
    步骤3-3、经过所述变化率控制方法平滑后的风储功率波动率
    Figure PCTCN2017082564-appb-100024
    的公式为：
    Figure PCTCN2017082564-appb-100025
    Figure PCTCN2017082564-appb-100026
    Figure PCTCN2017082564-appb-100027
    T＝nΔt    (14)
    式中：
    Figure PCTCN2017082564-appb-100028
    和
    Figure PCTCN2017082564-appb-100029
    是观测时间T内的风储功率最大值和最小值；
    变化率控制功率波动率方法的目标是控制平滑后的风储功率波动率在观测时间内小于给定目标值：
    Figure PCTCN2017082564-appb-100030
    式中：
    Figure PCTCN2017082564-appb-100031
    是给定观测时间T内风储功率波动率的控制目标；
    步骤3-4、所述储能***功率的计算公式为：
    PBESS(t)＝Phybrid(t)-Pwp(t)    (16)
    所述储能***的荷电状态SOC的计算公式如下：
    如果PBESS(t)＞0，储能***放电，SOC减小，
    Figure PCTCN2017082564-appb-100032
    如果PBESS(t)＜0，储能***充电，SOC增大，
    Figure PCTCN2017082564-appb-100033
    式中：Wbat是储能***容量。
  5. 根据权利要求1所述控制方法，其中，所述步骤(4)中，所述初始化训练评价模块和执行模块包括：初始化设置折扣因子α、执行模块学习率la和评价模块学习率lc、执行模块权值Wa和评价模块权值Wc，最大循环次数，以及训练模块的期望误差，其中权值的初始值设置为(-1,+1)间的随机值，学习率和折扣因子依据控制效果要求进行选取和调整。
  6. 根据权利要求1所述控制方法，其中，所述步骤(5)包括如下步骤：
    步骤5-1、判断风储功率波动率
    Figure PCTCN2017082564-appb-100034
    是否在约束条件
    Figure PCTCN2017082564-appb-100035
    内，若是则储能***不动作，不对储能功率进行修正；否则进行下一步的自适应动态规划模块训练，寻求最优储能功率修正值；
    步骤5-2、将被控对象状态和所述控制策略作为评价模块的输入，训练评价模块，更新评价模块的权值，输出代价函数；
    步骤5-3、将被控对象状态包括风储功率波动率
    Figure PCTCN2017082564-appb-100036
    和储能功率PBESS(t)作为执行模块的输入，训练执行模块，更新执行模块的权值，输出为控制策略，即储能功率的修正值ΔPBESS(t)；
    步骤5-4、保存此时刻的控制策略，并计算下一时刻被控对象状态，t＝t+1，重复步骤5-1到步骤5-3，直到控制过程结束。
  7. 根据权利要求6所述控制方法，其中，所述步骤5-1中，所述判断风储功率波动率
    Figure PCTCN2017082564-appb-100037
    的约束如下：
    如果
    Figure PCTCN2017082564-appb-100038
    则储能***过度出力，需要进行反向修正：
    Figure PCTCN2017082564-appb-100039
    式中，ΔPBESS(t)为储能功率修正值
    如果
    Figure PCTCN2017082564-appb-100040
    储能***出力适宜，无需修正：
    Figure PCTCN2017082564-appb-100041
    式中：
    Figure PCTCN2017082564-appb-100042
    是在变化率控制方法的基础上，对储能功率进行自适应动态规划调节后的储能***的功率。
  8. 根据权利要求6所述控制方法，其中，所述步骤5-2包括如下步骤：
    步骤5-2-1、将所述被控对象状态即风储功率波动率
    Figure PCTCN2017082564-appb-100043
    和所述控制策略即储能功率修正值ΔPBESS(t)进行归一化处理为[-1，+1]之间；
    步骤5-2-2、将所述被控对象状态即风储功率波动率
    Figure PCTCN2017082564-appb-100044
    和所述控制策略即储能功率修正值ΔPBESS(t)送入所述评价模块作为输入，计算所述评价模块的输出Jc WPBESS(t)代价函数，构造目标函数Echybrid(t)训练所述评价模块，并根据评价模块的权值更新式，更新评价模块神经网络的权值，计算公式如下：
    Figure PCTCN2017082564-appb-100045
    Figure PCTCN2017082564-appb-100046
    Figure PCTCN2017082564-appb-100047
    式中：代价函数Jchybrid(t)为评价模块的输出，效用函数U(t)是关于
    Figure PCTCN2017082564-appb-100048
    ΔPBESS(t),t的函数，根据控制目标进行定义，βc是折扣因子；
    步骤5-2-3、所述评价模块的训练采用梯度下降法或粒子群优化算法以使目标函数Echybrid(t)最小化为目标来更新评价模块神经网络的权值Wc，当目标函数Echybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
  9. 根据权利要求6所述控制方法，其中，所述步骤5-3包括如下步骤：
    步骤5-3-1、将被控对象状态包括风储功率波动率
    Figure PCTCN2017082564-appb-100049
    和储能功率PBESS(t)作为执行模块的输入，训练执行模块；
    步骤5-3-2、通过最小化评价模块输出Jchybrid(t)来调整控制策略即储能功 率修正值ΔPBESS(t)，公式如下：
    Figure PCTCN2017082564-appb-100050
    Figure PCTCN2017082564-appb-100051
    式中：控制策略ΔPBESS(t)是执行模块的输出，用来调整储能功率在合理范围内变化，以减小储能***SOC波动范围，u表示控制策略ΔPBESS(t)是关于
    Figure PCTCN2017082564-appb-100052
    的函数；
    步骤5-3-3、所述执行模块的训练采用梯度下降法或粒子群优化算法以最小化目标函数Eahybrid(t)为目标来更新执行模块神经网络的权值Wa，当目标函数Eahybrid(t)减小到设定的误差值或迭代次数达到最大时，训练完成。
  10. 一种大规模储能电站自适应动态规划的控制***，其中，所述***包括参数初始化模块、数据采集和计算模块、执行模块、评价模块及输出模块；
    所述参数初始化模块，配置为设置自适应动态规划控制***的结构和控制目标参数，并将所述目标参数传送到所述数据采集和计算模块；
    所述数据采集和计算模块，用于根据所述目标参数计算风储功率波动率，并将所述风储功率波动率传送到所述执行模块和评价模块；
    所述执行模块，配置为根据所述风储功率波动率得到储能功率修正值，并将所述储能功率修正值传送到所述数据采集和计算模块、评价模块及输出模；
    所述评价模块，配置为根据所述风储功率波动率得出代价函数，并将其传送到执行模块；
    所述输出模块，配置为输出所述各个时刻的控制策略，平滑后的风储功率波动率，储能功率和储能***荷电状态SOC。
  11. 一种大规模储能电站自适应动态规划的控制方法，所述方法包括 如下步骤：
    计算当前时刻t的原始风电功率波动率
    Figure PCTCN2017082564-appb-100053
    根据变化率控制策略对原始风电功率进行平滑，计算平滑后的风储功率波动率
    Figure PCTCN2017082564-appb-100054
    并获得所储能***功率PBESS(t)，储能***的荷电状态SOC；
    判断所述所述PBESS(t)及所述SOC是否在控制目标参数对应的约束范围内；
    当所述
    Figure PCTCN2017082564-appb-100056
    所述PBESS(t)及所述SOC不在所述约束范围内时，将所述
    Figure PCTCN2017082564-appb-100057
    输入到当前的执行模块；
    当前的所述执行模块基于所述
    Figure PCTCN2017082564-appb-100058
    输出控制策略，其中，所述控制策略用于控制所述大规模储能电站的储能***的充放电功率；
    将所述
    Figure PCTCN2017082564-appb-100059
    和当前控制策略输入到当前评价模块；
    所述评价模块基于所述
    Figure PCTCN2017082564-appb-100060
    及所述当前控制策略时，输出代价函数；
    基于所述
    Figure PCTCN2017082564-appb-100061
    及所述代价函数，构建训练所述评价模块的目标函数；
    以最小化所述目标函数为训练目的，依据所述目标函数训练所述评价模块；
    以最小化所述代价函数为目的，依据所述代价函数训练所述执行模块；其中，重新训练的所述评价模块和执行模块，用于下一时刻的所述控制策略的输出。
  12. 根据权利要求11所述的方法，其中，
    所述控制策略包括：储能功率修正值。
  13. 一种计算机存储介质，所述计算机存储介质中存储有计算机可执行指令，所述计算机可执行指令用于执行权利要求1至9及11至12任一 项所述的方法。
PCT/CN2017/082564 2016-04-28 2017-04-28 储能电站自适应动态规划的控制方法、***和存储介质 WO2017186178A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US16/095,844 US11326579B2 (en) 2016-04-28 2017-04-28 Adaptive dynamic planning control method and system for energy storage station, and storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201610278732.6 2016-04-28
CN201610278732.6A CN105846461B (zh) 2016-04-28 2016-04-28 一种大规模储能电站自适应动态规划的控制方法和***

Publications (1)

Publication Number Publication Date
WO2017186178A1 true WO2017186178A1 (zh) 2017-11-02

Family

ID=56590352

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2017/082564 WO2017186178A1 (zh) 2016-04-28 2017-04-28 储能电站自适应动态规划的控制方法、***和存储介质

Country Status (3)

Country Link
US (1) US11326579B2 (zh)
CN (1) CN105846461B (zh)
WO (1) WO2017186178A1 (zh)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111162551A (zh) * 2020-01-15 2020-05-15 国网内蒙古东部电力有限公司 一种基于风电功率超短期预测的蓄电池充放电控制方法
CN114611338A (zh) * 2022-05-11 2022-06-10 国网江西省电力有限公司电力科学研究院 一种储能电站选址定容方法及***
CN114614482A (zh) * 2022-05-12 2022-06-10 山东理工大学 基于连续时间尺度的虚拟储能提升配电***韧性的方法

Families Citing this family (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN105846461B (zh) 2016-04-28 2022-01-28 中国电力科学研究院 一种大规模储能电站自适应动态规划的控制方法和***
CN106230024B (zh) * 2016-08-22 2018-10-02 张雪敏 含双馈风电场的电力***停电风险计算方法
CN108270228B (zh) * 2017-01-03 2020-11-06 国家能源投资集团有限责任公司 储能***的控制方法、装置和储能***以及电力***
EP3635246B1 (en) * 2017-06-07 2023-07-19 Vestas Wind Systems A/S Adaptive estimation of available power for wind turbine
CN107147115B (zh) * 2017-06-16 2020-09-11 北京邮电大学 一种基于双时间尺度的用户侧能源管理方法
CN107492901B (zh) * 2017-08-29 2020-04-07 广东电网有限责任公司电力科学研究院 一种分布式储能***实时优化方法及装置
CN108376991B (zh) * 2018-02-09 2022-07-22 中国电力科学研究院有限公司 一种新能源电站储能***的综合能量管理方法及***
CN109213104B (zh) * 2018-09-19 2020-09-18 四川大学 基于启发式动态规划的储能***的调度方法及调度***
CN109725534A (zh) * 2018-12-29 2019-05-07 云南电网有限责任公司电力科学研究院 基于mmc的statcom控制器自适应动态规划方法
US11579680B2 (en) * 2019-02-01 2023-02-14 Alibaba Group Holding Limited Methods and devices for power management based on synthetic machine learning benchmarks
CN109884886B (zh) * 2019-03-29 2021-09-28 大连海事大学 一种基于宽度学习的船舶运动无模型自适应最优控制方法
CN110165714B (zh) * 2019-05-30 2021-01-26 广州水沐青华科技有限公司 基于极限动态规划算法的微电网一体化调度与控制方法、计算机可读存储介质
CN110429652B (zh) * 2019-08-28 2021-03-26 广西大学 一种可拓展深宽度自适应动态规划的智能发电控制方法
CN114746626A (zh) * 2019-12-16 2022-07-12 通用电气公司 用于维持基于蒸汽的电厂中的电力连续性的***和方法
CN111725827A (zh) * 2020-07-24 2020-09-29 山东电力交易中心有限公司 一种基于荷电状态自调节的储能平滑风电波动的控制方法
CN112861376B (zh) * 2021-03-10 2023-07-14 广东电网有限责任公司 一种基于单元调度模型的评估方法及装置
CN113241788B (zh) * 2021-06-04 2022-11-15 华能陇东能源有限责任公司 风电***异步采样实时削峰填谷控制方法、装置及存储介质
CN113312780A (zh) * 2021-06-07 2021-08-27 杭州市电力设计院有限公司余杭分公司 一种储能站规划方法、装置及设备
CN113552802B (zh) * 2021-07-22 2022-05-24 泰铂（上海）环保科技股份有限公司 重卡智能空调控制方法及***
CN113885360B (zh) * 2021-09-29 2024-04-16 北京金山云网络技术有限公司 智慧楼宇的电热水器控制方法、装置、可读存储介质及***
CN114421501B (zh) * 2022-01-07 2022-12-06 国网湖北省电力有限公司黄冈供电公司 用于分布式储能供电的自适应控制***参数确定方法
CN115483691B (zh) * 2022-11-15 2023-02-28 华北电力大学 储能协调漂浮式风电机组的控制方法及装置
CN116572791B (zh) * 2023-05-10 2023-12-19 淮阴工学院 一种电动汽车用自适应智能快充充电装置
CN116542882B (zh) * 2023-07-06 2023-09-19 浙江大学 光伏发电平滑方法、***和存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1022838A2 (en) * 1999-01-22 2000-07-26 Hitachi, Ltd. Electric power variation compensating device
CN102738817A (zh) * 2012-06-14 2012-10-17 华北电力大学 利用电池储能***对风电功率进行自适应平滑处理的方法
CN103956758A (zh) * 2014-05-20 2014-07-30 电子科技大学 一种风储***中储能soc优化控制方法
CN104022503A (zh) * 2014-06-18 2014-09-03 中国科学院自动化研究所 一种带有储能设备的智能微电网电能优化控制方法
WO2015054878A1 (zh) * 2013-10-18 2015-04-23 中国电力科学研究院 基于变化率控制储能电站平滑风光发电波动的方法及***
CN105846461A (zh) * 2016-04-28 2016-08-10 中国电力科学研究院 一种大规模储能电站自适应动态规划的控制方法和***

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8860236B2 (en) * 2009-10-19 2014-10-14 Uwm Research Foundation, Inc. Wind energy power conversion system reducing gearbox stress and improving power stability
JP5793572B2 (ja) * 2011-09-13 2015-10-14 日本風力開発株式会社 ピークカット制御装置
CN103560533B (zh) * 2012-04-18 2015-09-09 中国电力科学研究院 基于变化率控制储能电站平滑风光发电波动的方法及***
CN103248065B (zh) * 2013-05-22 2015-01-14 中国南方电网有限责任公司调峰调频发电公司 一种风电场中电池储能***的充放电控制方法
CN104283224B (zh) * 2013-07-01 2016-09-21 国家电网公司 一种限制风电波动率的储能***平滑风电功率控制方法
US10079317B2 (en) * 2013-07-15 2018-09-18 Constantine Gonatas Device for smoothing fluctuations in renewable energy power production cause by dynamic environmental conditions

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1022838A2 (en) * 1999-01-22 2000-07-26 Hitachi, Ltd. Electric power variation compensating device
CN102738817A (zh) * 2012-06-14 2012-10-17 华北电力大学 利用电池储能***对风电功率进行自适应平滑处理的方法
WO2015054878A1 (zh) * 2013-10-18 2015-04-23 中国电力科学研究院 基于变化率控制储能电站平滑风光发电波动的方法及***
CN103956758A (zh) * 2014-05-20 2014-07-30 电子科技大学 一种风储***中储能soc优化控制方法
CN104022503A (zh) * 2014-06-18 2014-09-03 中国科学院自动化研究所 一种带有储能设备的智能微电网电能优化控制方法
CN105846461A (zh) * 2016-04-28 2016-08-10 中国电力科学研究院 一种大规模储能电站自适应动态规划的控制方法和***

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
LI, XIANGJUN ET AL.: "Optimal Control Method of Energy Storage System Based on Adaptive Dynamic Programming", POWER SYSTEM TECHNOLOGY, vol. 40, no. 5, 31 May 2016 (2016-05-31), pages 1355 - 1362, ISSN: 1000-3673 *

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111162551A (zh) * 2020-01-15 2020-05-15 国网内蒙古东部电力有限公司 一种基于风电功率超短期预测的蓄电池充放电控制方法
CN114611338A (zh) * 2022-05-11 2022-06-10 国网江西省电力有限公司电力科学研究院 一种储能电站选址定容方法及***
CN114614482A (zh) * 2022-05-12 2022-06-10 山东理工大学 基于连续时间尺度的虚拟储能提升配电***韧性的方法
CN114614482B (zh) * 2022-05-12 2022-08-05 山东理工大学 基于连续时间尺度的虚拟储能提升配电***韧性的方法

Also Published As

Publication number Publication date
US20200191117A1 (en) 2020-06-18
CN105846461A (zh) 2016-08-10
CN105846461B (zh) 2022-01-28
US11326579B2 (en) 2022-05-10

Similar Documents

Publication Publication Date Title
WO2017186178A1 (zh) 储能电站自适应动态规划的控制方法、***和存储介质
WO2021164112A1 (zh) 风电场优化配置储能作为黑启动电源的频率控制方法及***
WO2017000853A1 (zh) 主动配电网多时间尺度协调优化调度方法和存储介质
WO2021114849A1 (zh) 增加新能源发电波动的海岛电网储能***分层控制方法
CN107301472B (zh) 基于场景分析法和电压调节策略的分布式光伏规划方法
CN105162149B (zh) 基于模糊自适应调节的光储***跟踪发电计划出力方法
WO2017161785A1 (zh) 基于储能运行状态的光伏功率稳定输出控制方法
CN104779631B (zh) 基于风电预测功率的储能跟踪风电出力计划方法及其***
CN111697597B (zh) 一种基于粒子群算法的火储联合agc调频控制方法
CN114362196B (zh) 一种多时间尺度主动配电网电压控制方法
CN110518643B (zh) 一种储能联合火电机组参与agc调频的控制方法及装置
CN107994595A (zh) 一种削峰填谷控制方法和***以及应用该控制方法的***
CN105896575B (zh) 基于自适应动态规划的百兆瓦储能功率控制方法及***
CN112398115B (zh) 一种基于改进模型预测控制的多时间尺度火电-光伏-抽水蓄能联合优化调度方案
CN107492901B (zh) 一种分布式储能***实时优化方法及装置
CN110137981B (zh) 一种基于一致性算法的分布式储能聚合器agc方法
CN112103980A (zh) 一种联合火电机组agc调频的混合储能***能量管理方法
CN104967149A (zh) 一种微电网风光储模型预测控制方法
CN110165698B (zh) 一种实现前瞻性误差资产转换的风电场平滑并网方法
CN105680478A (zh) 基于超短期预测的储能***平滑新能源发电控制方法
Han et al. Optimization control and economic evaluation of energy storage combined thermal power participating in frequency regulation based on multivariable fuzzy double-layer optimization
Yang et al. Tracking-dispatch of a combined wind-storage system based on model predictive control and two-layer fuzzy control strategy
CN103679284A (zh) 一种消纳风电接入的定区间滚动调度方法
CN114865649B (zh) 一种风光储一体化场站无功调节方法、装置及电子设备
CN107453366B (zh) 计及风电决策风险的含upfc多目标最优潮流计算方法

Legal Events

Date Code Title Description
NENP Non-entry into the national phase

Ref country code: DE

121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 17788830

Country of ref document: EP

Kind code of ref document: A1

122 Ep: pct application non-entry in european phase

Ref document number: 17788830

Country of ref document: EP

Kind code of ref document: A1