CN110165714B - Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium - Google Patents
Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium Download PDFInfo
- Publication number
- CN110165714B CN110165714B CN201910464429.9A CN201910464429A CN110165714B CN 110165714 B CN110165714 B CN 110165714B CN 201910464429 A CN201910464429 A CN 201910464429A CN 110165714 B CN110165714 B CN 110165714B
- Authority
- CN
- China
- Prior art keywords
- micro
- power
- microgrid
- module
- control
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- 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
- H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
-
- 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]
Abstract
The invention relates to the field of power system optimization control technology application, and provides a micro-grid integrated scheduling and control method based on an extreme dynamic programming algorithm and a computer readable storage medium. The micro-grid integrated scheduling and control method based on the extreme dynamic programming algorithm adopts the dynamic programming algorithm, droop control and economic scheduling are used in a matched mode, the limitation of a trial and error mechanism of a traditional intelligent algorithm is avoided, and the speed of algorithm pre-learning is greatly improved by adding the extreme learning machine. In the operation process of the algorithm, the prediction capability and the decision-making capability of the algorithm to the system are continuously enhanced through online learning fine tuning, so that a better control effect is obtained, the integrated scheduling and control of the micro-grid can be quickly and efficiently realized, and the accuracy and the learning speed are considered.
Description
Technical Field
The invention relates to the field of power system optimization control technology application, in particular to a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium and is executed by a controller to realize a micro-grid integrated scheduling and control method based on an extreme dynamic programming algorithm.
Background
A Micro Grid (MG) refers to a small power generation and distribution system composed of various distributed power sources, an energy storage device, an energy conversion device, a monitoring/protection device, and the like. With the background of the increasing maturity of intelligent control technology and energy management technology, the construction of micro-grids has gradually become a necessary trend of distributed power generation network structures. In addition, a large amount of renewable energy power sources in the microgrid have large intermittency and uncertainty, so that the power of the microgrid is often in an unstable state, and power balance adjustment is needed. In a traditional mode, a power balance adjustment method is mainly used for distributing power generation control instructions obtained based on a PID control algorithm to each micro power supply through an optimization algorithm, wherein common optimization algorithms include a classical Genetic Algorithm (GA), a Particle Swarm Optimization (PSO), and the like. In consideration of power generation cost control, a microgrid control system usually realizes a layered control scheme on the basis of power balance adjustment, and a three-layer control combination strategy of droop control, automatic power generation control and economic dispatching is formed.
However, the control combination strategy of the microgrid system mainly aims at reducing frequency deviation, and the lack of cooperation among the hierarchical control strategy levels often leads to incomplete consideration of economic benefits, so that distributed power generation is uneconomical in economic dispatching after droop control and automatic power generation control. For this reason, it is necessary to centralize the three-layer control of the microgrid to a unified control platform, i.e. to form an integrated control system. The micro-grid integrated control needs a multi-condition constrained multi-target decision model, and the traditional multi-target optimization method which is applied more at present mainly adopts methods such as weighting, constraint and the like to convert a multi-target problem into a single-target optimization problem for simplifying and solving, so that errors and limitations are large.
Disclosure of Invention
The invention aims to: the shortcomings in the prior art are avoided, the dynamic limit planning algorithm which is short in time consumption and high in precision and is applied to micro-grid control is provided, and integrated scheduling and control of the micro-grid can be achieved.
The idea of the invention is as follows:
the invention discloses an Adaptive Dynamic Programming (ADP) algorithm, which comprises an execution module, an evaluation module and a model module, belongs to an optimal control solving method, and realizes the micro-grid integrated scheduling problem of large-time-scale and large-scale units on the basis of the ADP algorithm. The optimal decision process of the ADP algorithm is a process of performing online learning by taking a minimized cost function as a target through an ADP algorithm logic, wherein an execution module outputs control actions according to a decision strategy according to a current state, a model module estimates a state at the next moment according to the current state and the control actions output by the execution module, and an evaluation module estimates the cost function according to the state at the next moment and feeds back the cost function to the execution module. The specific working principle of the ADP algorithm in the integrated scheduling and control of the microgrid system is as follows: the execution module provides a decision action according to the load of the micro-grid system and the initial state of the output power of each controllable micro-power supply, the decision action is used for controlling the output power of each controllable micro-power supply in the micro-grid, and the decision action is provided for the model module and the evaluation module; the model module calculates the state of the next moment through a system state equation according to the initial state and the decision action from the execution module, and the state of the next moment is provided for the evaluation module; the evaluation module evaluates the current decision-making action according to the decision-making action from the execution module and the next time state from the model module, calculates a cost function and outputs a penalty function value to the execution module according to the cost function; and the execution module performs feedback adjustment according to the received penalty function value, and minimizes the cost function in a function approximation mode, so that an optimized decision strategy is obtained. Compared with other existing optimal control methods, the ADP algorithm has a unique algorithm and structure, can be used for off-line learning control firstly and then continuously learning on line in the operation process to obtain knowledge, does not depend on a mathematical model of a controlled object, can output multiple control instructions according to input and control targets, can better process multilayer control integration tasks in the operation process of the micro-grid, and is suitable for the micro-grid system. Moreover, the ADP algorithm adopts a successive approximation method to solve the value function of each state in each stage, so that the problem of hierarchy imbalance in the dynamic planning process can be effectively avoided. Moreover, different from the conventional ADP algorithm, an artificial neural network such as a BP algorithm (Back Propagation) is used in the calculation of the value function, in order to ensure the accuracy of the algorithm, a large amount of learning calculation is required before the actual use, which takes a long time, and the accuracy and the learning speed cannot be considered at the same time. The inventor thinks that in the process of realizing optimal control by adopting the ADP algorithm, an Extreme Learning Machine (ELM) is applied to each module of the ADP algorithm as a solving method, so that the time for Learning calculation in the ADP algorithm is reduced, and the overall calculation process of the algorithm is short in time and high in precision.
The purpose of the invention is realized by the following technical scheme:
the method for integrally scheduling and controlling the microgrid based on the extreme dynamic programming algorithm comprises the following steps:
the method comprises the steps of data acquisition, namely acquiring matched power generation power, cost and load data of a micro-grid controllable micro-power source, and storing the data into a training data set and a test sample set;
a controller construction step, namely establishing a controller model based on a self-adaptive dynamic programming algorithm, wherein the controller model comprises an execution module, an evaluation module and a model module;
module learning step: according to the extreme learning machine, training each module of the controller model by adopting a training data set, and optimizing the trained extreme learning machine by adopting a test sample set;
a microgrid control step: the method comprises the following steps that an evaluation module and an execution module respectively acquire input states of a micro-grid system, and the execution module obtains control actions of each controllable micro-power source through a power optimal control strategy of an extreme learning algorithm according to the input states and sends the control actions to a model module and the evaluation module; the model module redistributes the generated power among all controllable micro power sources by adopting an economic dispatching condition as an optimal control strategy to obtain the state of the micro power grid system at the next moment and output the state to the evaluation module; the evaluation module obtains a penalty function value according to the received control action and the state at the next moment from the model module and outputs the penalty function value to the execution module; and the execution module adjusts the output action according to the penalty function value obtained from the evaluation module and outputs the action to each controllable micro power supply.
Preferably, in the controller constructing step, the controller model further includes a prediction module, the prediction module outputs a next-time state prediction value according to an input state of the microgrid system in the microgrid control step and transmits the next-time state prediction value to the execution module, and the execution module combines the input state and the next-time state prediction value to serve as a current input state, so as to obtain each controllable microgrid control action.
Preferably, the execution module combines the input state and the state prediction value at the next moment to refer to: and averaging the state values of the two.
Preferably, in the microgrid control step, the input state of the microgrid system is a total load.
Preferably, in the microgrid control step, the input and output actions of the execution module are the power of each controllable microgrid.
Preferably, in the microgrid control step, in the evaluation module, CDE,iRepresenting the power generation cost of the ith first type power source; cFC,jRepresenting the power generation cost of the jth second-type power supply; piRepresenting the actual output active power of the ith controllable micro power supply; a isi、biAnd ciA cost factor representing the ith first type of power supply; cfuelRepresents an additional cost of the second type of power supply; etajRepresenting the power generation efficiency of the jth second-type power supply; the first constraint represents the power constraint of the controllable micro-power supply, Pi maxAnd Pi minRespectively representing the upper and lower output limits of the ith micro power supply; the second constraint represents the system power balance constraint, PdRepresenting the power demand of the microgrid;
the cost of power generation for each micro power source is expressed as:
according to the minimum power generation cost minC of the micro-grid systemcostOutputs a penalty function value according to the change condition of the input signal.
Preferably, in the microgrid control step, the model module calculates power distribution of the controllable microgrid by using droop control, and outputs a next-time state of the microgrid system.
Preferably, according to microgrid power balance constraints
Droop control is expressed as:
f＝f0+mpi·(Prefi-Pi)，
wherein Pi denotes the actual output active power of the ith controllable micro power supply, PdRepresenting the power demand, P, of the microgridrefiRepresenting the reference output active power of the ith controllable micro power supply;
when the total load change is Δ P, the shared power of the ith distributed power supply is:
wherein, Δ Pi＝Prefi-PiIndicating the shared power of the ith distributed power source,
representing total power demand fluctuations of the microgrid;
and the equivalent droop coefficient of the whole microgrid is represented.
Preferably, said PrefiPower allocation value according to economic dispatch
Set up and satisfy
The sum equals the sum of the load demands of the whole microgrid, i.e.
The storage medium stores an executable computer program, and the computer program can realize the integrated micro-grid dispatching and controlling method based on the extreme dynamic programming algorithm when being executed by the controller.
The invention has the beneficial effects that: the micro-grid integrated dispatching and controlling method based on the extreme dynamic programming algorithm realizes the function of matching the droop control and the economic dispatching, and the effect of the method is superior to the effect of matching. The dynamic programming algorithm is adopted, so that the limitation of a trial and error mechanism of the traditional intelligent algorithm is avoided, the phenomenon that the traditional intelligent algorithm falls into a local optimal point is avoided, and the speed of algorithm pre-learning is greatly improved by adding the extreme learning machine. In the running process of the algorithm, the prediction capability and decision capability of the algorithm on the system are continuously enhanced through online learning fine tuning, so that a better control effect is obtained. The micro-grid integrated dispatching and controlling method can quickly and efficiently realize the integrated dispatching and controlling of the micro-grid, and give consideration to both accuracy and learning speed.
Drawings
The invention is further illustrated by means of the attached drawings, but the embodiments in the drawings do not constitute any limitation to the invention, and for a person skilled in the art, other drawings can be obtained on the basis of the following drawings without inventive effort.
Fig. 1 is an overall flow chart of the integrated micro-grid scheduling and controlling method based on the extreme dynamic programming algorithm.
Fig. 2 is a schematic diagram of a microgrid structure of the microgrid integrated scheduling and controlling method based on the extreme dynamic programming algorithm.
Fig. 3 is a schematic structural diagram of a controller agent of the microgrid integrated scheduling and control method based on the extreme dynamic programming algorithm.
Fig. 4 is a schematic diagram of a logic structure of an extreme learning machine of the micro-grid integrated scheduling and control method based on the extreme dynamic programming algorithm.
FIG. 5 is a schematic diagram of a typical single hidden layer feedforward neural network.
Fig. 6 is a comparison graph of the control effect of the integrated micro-grid scheduling and controlling method based on the extreme dynamic programming algorithm.
Detailed Description
The invention is further described with reference to the following examples.
The base limitThe microgrid integrated scheduling and control method based on the dynamic programming algorithm is realized based on a controller agent, and as shown in fig. 2, the microgrid comprises the controller agent and a microgrid system respectively connected to the controller agent. The micro-power system comprises a controllable micro-power DG which is independently communicated with the intelligent controller body respectively1、DG2、DG3、DG4、DG5And uncontrollable micro power sources such as wind power generation, photovoltaic power generation and the like are taken as boundary conditions to account for load fluctuation. In addition, another control device communicated with the controller intelligence body of the microgrid is arranged at a switch where the microgrid is connected with the main power grid, and the algorithm model of the microgrid can be expanded to be connected with the main power grid for operation, so that each module of the dynamic planning algorithm has stronger convergence and rapidness.
As shown in fig. 2 and 3, the microgrid integrated scheduling and control method based on the extreme dynamic programming algorithm adds a prediction module on the basis of a conventional ADP algorithm, and each module adopts an extreme learning machine as a computing kernel to obtain a new Extreme Dynamic Programming (EDP) algorithm. The specific operation steps of the EDP algorithm are shown in fig. 1, and under the normal operation state of the microgrid, microgrid load, controllable microgrid power generation power and cost data are obtained through an intelligent acquisition terminal. The data are stored in a server, a training data set and a sample data set of the extreme learning machine in the extreme dynamic programming algorithm are established by using the data, and the extreme learning machine of each module is trained and optimized in an algorithm controller (see figure 4) consisting of an execution module, an evaluation module, a model module and a prediction module, so that the extreme dynamic programming algorithm controller containing the trained extreme learning machine is obtained. In the actual operation process of the algorithm controller, an action matrix formed by the power of each controllable micro power supply of the micro power grid system is used as an input action and is respectively sent to the evaluation module and the execution module, and the prediction module and the execution module respectively acquire the total load of the micro power grid system as the current input state of the algorithm controller. And the prediction module outputs a state prediction value at the next moment according to the current input state and transmits the state prediction value to the execution module. The execution module combines the current input state with the state predicted value at the next moment and takes the combined value as state input, then obtains output action according to the penalty function value obtained from the evaluation module and the power optimal control strategy of the extreme learning algorithm, and then transmits the output action to a droop control curve link in each inverter of the controllable micro power supply of the micro power grid to realize the control and regulation of the power of the micro power grid system; on the other hand, the execution module sends the output action to the model module, the model module redistributes the power generation power among all the controllable micro power sources by adopting the economic dispatching condition as an optimal control strategy to obtain the next-time state of the micro-grid system and outputs the next-time state to the evaluation module, and the evaluation module obtains a punishment function value according to the input action received by the round and the next-time state fed back by the model module and outputs the punishment function value to the execution module.
Specifically, the intelligent acquisition terminal of the microgrid system can acquire data such as voltage, current, power and power generation cost of various controllable micro power supplies of the microgrid at high frequency, and has the capacity of uploading mass data to a server for storage.
Specifically, in the limit evaluation module, the power generation cost of each micro power supply is expressed as:
wherein, CDE,iRepresents the power generation cost of the ith diesel engine; cFC,jRepresents the power generation cost of the jth gas turbine/fuel cell; piRepresenting the actual output active power of the ith controllable micro power supply; pDE,iRepresenting the actual output active power of the ith first type power supply; pFC,jRepresenting the actual output active power of the jth second type power supply; a isi、biAnd ciRepresenting the cost coefficient of the ith diesel engine; cfuelRepresents the fuel cost of the gas turbine/fuel cell; etajRepresents the power generation efficiency of the jth gas turbine/fuel cell; the first constraint represents the power constraint of the controllable micro-power supply, Pi maxAnd Pi minRespectively representThe upper and lower output limits of the ith distributed power supply; the second constraint represents the system power balance constraint, PdRepresenting the power demand of the microgrid. The solving process of the limit evaluation module is the minimum power generation cost minCcostThe power matching process of (1).
In the limit model module, droop control is adopted to distribute the power of the controllable micro power supply and output the next time state of the micro power grid system, and under the stable state of the micro power grid, power balance constraint is carried out
The droop control will redistribute the generated power between the various distributed power sources, which is satisfied from time to time, i.e. when the load fluctuates.
Specifically, droop control is expressed as:
f＝f0+mpi·(Prefi-Pi) (2)
wherein, PrefiThe output active power of the ith controllable micro power supply is represented, and when the total load change is delta P, the shared power of the ith distributed power supply is as follows:
wherein, Δ Pi＝Prefi-PiRepresenting the shared power of the ith distributed power supply;
representing total power demand fluctuations of the microgrid;
and the equivalent droop coefficient of the whole microgrid is represented.
If P in the formula (1)refiPower allocation value according to economic dispatch
Set up and satisfy
The sum of the load demands of the whole microgrid is equal to
Is available in a stable state
From this combination (3) can be obtained
F ═ f available from combined formula (2)0. From this, it can be seen that by setting PrefiHas a value of
The combination of economic dispatch and automatic power generation control can be realized on the same time scale.
In practice, it is assumed that one includes two distributed power Sources (DGs)iAnd DGj) Under the condition of not counting loss, P is the micro-grid systemrefiAnd PrefjAre all set to be 0, the frequency of the micro-grid system is f, and the output power of the two distributed power supplies is PiAnd PjAt this time, the power balance P is satisfiedd＝Pi+Pj. And assume that the system load is PdUnder the initial condition, the economic dispatch allocation values of the two power supplies are respectively Pi *And
at this time should still satisfy
This time game
And
power of two power supplies PiAnd PjWill eventually converge to respectively at steady state
And
at this time, the frequency of the system is restored to the rated frequency f0。
Using ADP-core algorithms in limit execution blocks
The first expression represents the state recursion of the system, namely in the k stage, when the system is in the x (k) state, the system will run to the x (k +1) state after the u (k) strategy is adopted. Strategy selection method of the second expression algorithm, wherein u*(k) The optimal decision of the k decision stage is obtained; l (x (k), u (k)) represents the immediate penalty function for taking u (k) decisions at the x (k) state; j. the design is a square*(x (k +1)) represents the function of the state value at x (k + 1). The specific adjusting process of each controllable micro power supply frequency is the same as the droop control adjusting process.
As shown in fig. 3, in the integrated micro-grid scheduling and controlling method based on the extreme dynamic programming algorithm, the load of the micro-grid system is used as an initial state, the power of a controllable micro-power supply is used as an action input, an economic scheduling condition is used as an optimal control strategy condition of an ADP algorithm evaluation module, and a frequency adjusting mode of droop control is used as an optimal control strategy condition of an execution module, so that an optimal output action is obtained. And the controller intelligent agent is in communication connection with each distributed controllable micro power supply, and transmits the output optimal strategy to a droop control curve link in each micro power supply inverter to obtain the micro power grid integrated scheduling and control method based on the extreme dynamic programming algorithm, so that each controllable micro power supply can provide stable frequency together while the micro power grid meets economic conditions.
The extreme dynamic programming algorithm adopts a cost control condition to train an extreme learning algorithm, the power value of each controllable micro power supply and the load condition of the micro-grid are input, the electric power of the controllable micro power supply under the descending trend of each controllable micro power supply is output and used for regulating and controlling, the electric power is used as the output optimal regulation and control parameter value and is obtained after the ADP algorithm and the extreme learning machine are combined, the problem of long time of learning and training is avoided, and the problem that the algorithm structure cannot be updated in real time in the feedback process of the neural network is also avoided.
According to the micro-grid integrated scheduling and control method based on the Extreme dynamic programming algorithm, an Extreme Learning Machine (ELM) is applied to each module of an ADP algorithm to serve as a solving method. Specifically, the extreme learning machine is a machine learning algorithm designed for a feedforward neural network (feedforward neural network) as shown in fig. 5, and randomly assigns values to input feature vectors and weights from an input layer to a hidden layer, while the weights from the hidden layer to an output layer need to be obtained according to a least square method, that is, connection weights between the input layer and the hidden layer and a threshold of a neuron of the hidden layer are randomly generated, and only the number of neurons of the hidden layer needs to be set without adjustment in a training process, so that a unique optimal solution can be obtained. Compared with the traditional BP neural network algorithm, the ELM algorithm has the advantages of high learning speed and good generalization performance.
Specifically, the extreme learning machine is realized based on the following theory.
Theorem 1: given an arbitrary number Q of different samples (x)i,ti) Wherein, in the step (A),
xi＝[xi1,xi2,…,xin]T∈Rn，
ti＝[ti1,ti2,…,tim]∈Rm，
r → R is an infinitely differentiable activation function in an arbitrary interval, and w is assigned arbitrarily to a single hidden layer feedforward neural network with Q hidden layer neuronsi∈RnAnd biE.g. in case of R, its hidden layer output matrix H is invertible and has | | | H β -T' | | 0.
From the above theorem 1, if the number of hidden layer neurons is equal to the number of training set samples, the SLFN can approach the training samples with zero error for any w and b, i.e. the SLFN
Wherein the content of the first and second substances,
yj＝[y1j,y2j,…,ymj]T(j＝1,2,…,Q)。
however, when the number Q of training samples is large, the number K of hidden layer neurons is usually a smaller number than Q in order to reduce the amount of computation.
Theorem 2: given an arbitrary small error ε > 0 and an infinitely differentiable activation function g: R → R in the interval, there is always a K (K) containing<Q) SLFN of hidden layer neurons, assigning w at willi∈RnAnd biE is in the case of R, has | | | HN×MβM×m-T′||＜ε。
From theorem 2 above, the training error of SLFN approaches an arbitrary ε > 0, i.e.
Thus, when the activation function g (x) is infinitely differentiable, the parameters of the SLFN need not all be adjusted, w and b may be randomly selected before training and remain unchanged during training. And the connection weight beta of the hidden layer and the output layer can be obtained by solving the least square solution of the following equation set:
based on the theory, the extreme learning machine randomly generates the connection weight of the input layer and the hidden layer and the threshold of the neuron of the hidden layer, and only needs to set the number of the hidden neurons without adjustment in the training process to obtain a unique optimal solution, so that the module value functions are solved, and the adjustment of the power of each micro-power supply in the micro-grid is realized in the extreme dynamic programming algorithm.
The micro-grid integrated scheduling and control method of the extreme dynamic programming algorithm is verified through tests, the EDP algorithm, the ADP algorithm and the EDP algorithm without a prediction module are respectively adopted for frequency control, the control effect is shown in figure 6, and the EDP algorithm can guarantee extremely fast response speed compared with other control algorithms while guaranteeing control accuracy.
The micro-grid integrated dispatching and controlling method based on the extreme dynamic programming algorithm realizes the function of matching the droop control and the economic dispatching, and the effect of the method is superior to the effect of matching. The dynamic programming algorithm is adopted, so that the limitation of a trial and error mechanism of the traditional intelligent algorithm is avoided, the phenomenon that the traditional intelligent algorithm falls into a local optimal point is avoided, and the speed of algorithm pre-learning is greatly improved by adding the extreme learning machine. In the running process of the algorithm, the prediction capability and decision capability of the algorithm on the system are continuously enhanced through online learning fine tuning, so that a better control effect is obtained.
Finally, it should be noted that the above embodiments are only used for illustrating the technical solutions of the present invention, and not for limiting the protection scope of the present invention, although the present invention is described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that modifications or equivalent substitutions can be made on the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.
Claims (7)
1. The micro-grid integrated scheduling and control method based on the extreme dynamic programming algorithm is characterized by comprising the following steps of:
the method comprises the steps of data acquisition, namely acquiring matched power generation power, cost and load data of a micro-grid controllable micro-power source, and storing the data into a training data set and a test sample set;
a controller construction step, namely establishing a controller model based on a self-adaptive dynamic programming algorithm, wherein the controller model comprises an execution module, an evaluation module and a model module;
module learning step: according to the extreme learning machine, training each module of the controller model by adopting a training data set, and optimizing the trained extreme learning machine by adopting a test sample set;
a microgrid control step: the method comprises the following steps that an evaluation module and an execution module respectively acquire input states of a micro-grid system, and the execution module obtains control actions of each controllable micro-power source through a power optimal control strategy of an extreme learning algorithm according to the input states and sends the control actions to a model module and the evaluation module; the model module redistributes the generated power among all controllable micro power sources by adopting an economic dispatching condition as an optimal control strategy to obtain the state of the micro power grid system at the next moment and output the state to the evaluation module; the evaluation module obtains a penalty function value according to the received control action and the state at the next moment from the model module and outputs the penalty function value to the execution module; and the execution module adjusts the output action according to the penalty function value obtained from the evaluation module and outputs the action to each controllable micro power supply.
2. The integrated scheduling and control method of microgrid based on extreme dynamic programming algorithm according to claim 1, characterized in that in the controller construction step, the controller model further comprises a prediction module, the prediction module outputs a next-time state prediction value according to the input state of the microgrid system in the microgrid control step and transmits the next-time state prediction value to the execution module, and the execution module combines the input state and the next-time state prediction value as a current input state, thereby obtaining each controllable microgrid control action.
3. The integrated microgrid scheduling and control method based on a limit dynamic programming algorithm as claimed in claim 2, wherein the execution module combines the input state and the predicted value of the state at the next moment by: and averaging the state values of the two.
4. The integrated microgrid scheduling and control method based on a limit dynamic programming algorithm as claimed in claim 1, characterized in that in the microgrid control step, the input state of the microgrid system is the total load.
5. The integrated microgrid scheduling and control method based on a limit dynamic programming algorithm according to claim 1, characterized in that in the microgrid control step: cDE,iRepresenting the power generation cost of the ith first type power source; cFC,jRepresenting the power generation cost of the jth second-type power supply; piRepresenting the actual output active power of the ith controllable micro power supply; pDE,iRepresenting the actual output active power of the ith first type power supply; pFC,jRepresenting the actual output active power of the jth second type power supply; a isi、biAnd ciA cost factor representing the ith first type of power supply; cfuelRepresents an additional cost of the second type of power supply; etajRepresenting the power generation efficiency of the jth second-type power supply; the first constraint represents the power constraint of the controllable micro-power supply, Pi maxAnd Pi minRespectively representing the upper and lower output limits of the ith micro power supply; the second constraint represents the system power balance constraint, PdRepresenting the power demand of the microgrid;
the cost of power generation for each micro power source is expressed as:
according to the minimum power generation cost minC of the micro-grid systemcostOutputs a penalty function value according to the change condition of the input signal.
6. The integrated scheduling and control method for the microgrid based on the extreme dynamic programming algorithm of claim 1, wherein in the microgrid control step, the model module calculates the power distribution of the controllable microgrid by using droop control and outputs the next-time state of the microgrid system.
7. Computer-readable storage medium, in which a computer program is stored, which computer program, when being executed by a controller, is adapted to carry out the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910464429.9A CN110165714B (en) | 2019-05-30 | 2019-05-30 | Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910464429.9A CN110165714B (en) | 2019-05-30 | 2019-05-30 | Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110165714A CN110165714A (en) | 2019-08-23 |
| CN110165714B true CN110165714B (en) | 2021-01-26 |
Family
ID=67630233
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910464429.9A Active CN110165714B (en) | 2019-05-30 | 2019-05-30 | Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110165714B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110806463A (en) * | 2019-10-29 | 2020-02-18 | 首钢京唐钢铁联合有限责任公司 | Method and system for detecting atmosphere in annealing furnace |
| CN111242436B (en) * | 2020-01-03 | 2022-05-03 | 浙江工业大学 | Intelligent power grid real-time economic dispatching method based on distributed machine learning |
| CN112465289B (en) * | 2020-10-10 | 2024-04-19 | 浙江工业大学 | Smart grid economic dispatching method based on integrated neural network |
| CN112507614B (en) * | 2020-12-01 | 2021-09-07 | 广东电网有限责任公司中山供电局 | Comprehensive optimization method for power grid in distributed power supply high-permeability area |
| CN114397876A (en) * | 2022-01-24 | 2022-04-26 | 中国信息通信研究院 | Method and apparatus for evaluating performance of domain controller, electronic device, and storage medium |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000324694A (en) * | 1999-05-07 | 2000-11-24 | Koichi Tsuji | Method for deciding system operation procedures in power system |
| CN103501004A (en) * | 2013-10-25 | 2014-01-08 | 陕西省地方电力（集团）有限公司 | Operation control method and device for distribution network |
| CN104022503A (en) * | 2014-06-18 | 2014-09-03 | 中国科学院自动化研究所 | Electric-energy optimal control method for intelligent micro-grid with energy storage device |
| CN105203869A (en) * | 2015-09-06 | 2015-12-30 | 国网山东省电力公司烟台供电公司 | Microgrid island detection method based on extreme learning machine |
| CN105846461A (en) * | 2016-04-28 | 2016-08-10 | 中国电力科学研究院 | Self-adaptive dynamic planning control method and system for large-scale energy storage power station |
| CN105870942A (en) * | 2016-05-18 | 2016-08-17 | 中国电力科学研究院 | Primary frequency regulation additional learning control method based on approximate dynamic programming algorithm |
| CN107508275A (en) * | 2017-08-04 | 2017-12-22 | 华中科技大学 | A kind of DC micro-electric network control method and system based on adaptive Dynamic Programming |
| CN109449925A (en) * | 2018-10-29 | 2019-03-08 | 国网甘肃省电力公司 | A kind of adaptive dynamic programming method of multiple target joint optimal operation |
-
2019
- 2019-05-30 CN CN201910464429.9A patent/CN110165714B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2000324694A (en) * | 1999-05-07 | 2000-11-24 | Koichi Tsuji | Method for deciding system operation procedures in power system |
| CN103501004A (en) * | 2013-10-25 | 2014-01-08 | 陕西省地方电力（集团）有限公司 | Operation control method and device for distribution network |
| CN104022503A (en) * | 2014-06-18 | 2014-09-03 | 中国科学院自动化研究所 | Electric-energy optimal control method for intelligent micro-grid with energy storage device |
| CN105203869A (en) * | 2015-09-06 | 2015-12-30 | 国网山东省电力公司烟台供电公司 | Microgrid island detection method based on extreme learning machine |
| CN105846461A (en) * | 2016-04-28 | 2016-08-10 | 中国电力科学研究院 | Self-adaptive dynamic planning control method and system for large-scale energy storage power station |
| CN105870942A (en) * | 2016-05-18 | 2016-08-17 | 中国电力科学研究院 | Primary frequency regulation additional learning control method based on approximate dynamic programming algorithm |
| CN107508275A (en) * | 2017-08-04 | 2017-12-22 | 华中科技大学 | A kind of DC micro-electric network control method and system based on adaptive Dynamic Programming |
| CN109449925A (en) * | 2018-10-29 | 2019-03-08 | 国网甘肃省电力公司 | A kind of adaptive dynamic programming method of multiple target joint optimal operation |
Non-Patent Citations (5)
| Title |
|---|
| 含非可靠分布式电源的配电网孤岛划分;李滨等;《电力系统自动化》;20150425;第39卷(第8期);59-65 * |
| 基于动态规划的微电网动态经济调度;蒋一鎏等;《电气应用》;20161120;第35卷(第22期);67-72 * |
| 基于深度自适应动态规划的孤岛主动配电网发电控制与优化一体化算法;殷林飞等;《控制理论与应用》;20180215;第35卷(第2期);169-183 * |
| 基于自适应动态规划的储能系统优化控制方法;李相俊等;《电网技术》;20160505;第40卷(第5期);1355-1362 * |
| 综合能源微网系统的滚动优化经济调度;杜妍等;《电力系统及其自动化学报》;20171115;第29卷(第11期);20-25 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110165714A (en) | 2019-08-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110165714B (en) | Micro-grid integrated scheduling and control method based on extreme dynamic programming algorithm and computer readable storage medium | |
| CN110416998B (en) | Regional complex distribution network scheduling control management system based on virtual power plant | |
| CN111242443B (en) | Deep reinforcement learning-based economic dispatching method for virtual power plant in energy internet | |
| CN111697597B (en) | Fire storage combined AGC frequency modulation control method based on particle swarm optimization | |
| CN110429652B (en) | Intelligent power generation control method capable of expanding deep width self-adaptive dynamic planning | |
| CN110858718B (en) | Alternating current micro-grid distributed event-driven frequency control method considering economy | |
| CN109167347B (en) | Cloud-adaptive-particle-swarm-based multi-target electric vehicle charge-discharge optimization scheduling method | |
| CN110571865B (en) | Active power distribution network source load interaction decision method based on consistency algorithm | |
| CN113224769A (en) | Multi-time scale power distribution network voltage optimization method considering photovoltaic multi-state adjustment | |
| Cominesi et al. | A multi-layer control scheme for microgrid energy management | |
| CN114331059A (en) | Electricity-hydrogen complementary park multi-building energy supply system and coordinated scheduling method thereof | |
| CN113159366A (en) | Multi-time scale self-adaptive optimization scheduling method for multi-microgrid system | |
| CN113872213B (en) | Autonomous optimization control method and device for power distribution network voltage | |
| CN115795992A (en) | Park energy Internet online scheduling method based on virtual deduction of operation situation | |
| Ahmadi et al. | Performance of a smart microgrid with battery energy storage system's size and state of charge | |
| Yin et al. | Expandable deep width learning for voltage control of three-state energy model based smart grids containing flexible energy sources | |
| CN114330869A (en) | Multi-energy day-ahead scheduling optimization method for iron and steel enterprise considering multi-level indexes | |
| CN114400675B (en) | Active power distribution network voltage control method based on weight mean value deep double-Q network | |
| CN109245178B (en) | Wind power plant power cooperative scheduling based on distributed cooperative MPC | |
| CN110289643B (en) | Rejection depth differential dynamic planning real-time power generation scheduling and control algorithm | |
| CN116050461A (en) | Improved method for determining membership function of fuzzy control theory by using convolutional neural network | |
| CN115760178A (en) | Multi-region dynamic economic scheduling method with real-time power balance constraint | |
| CN114362169A (en) | Layered coordination regulation and control method considering marginal cost of light storage type virtual power plant | |
| CN114421479A (en) | Voltage control method for AC/DC micro-grid group cooperative mutual supply | |
| CN114384931A (en) | Unmanned aerial vehicle multi-target optimal control method and device based on strategy gradient |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information | ||
| CB02 | Change of applicant information |
Address after: Room 509, building D, 39 Ruihe Road, Huangpu District, Guangzhou City, Guangdong Province Applicant after: Guangzhou Shuimu Qinghua Technology Co.,Ltd. Address before: Room 936, Room 333 Jiufojian Road, Zhongxin Guangzhou Knowledge City, Guangzhou, Guangdong Province Applicant before: Guangzhou Shuimu Qinghua Technology Co.,Ltd. |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder | ||
| CP02 | Change in the address of a patent holder |
Address after: Room 604, Building 2, No. 5 Yunkai Road, Huangpu District, Guangzhou City, Guangdong Province, 510535 Patentee after: Guangzhou Shuimu Qinghua Technology Co.,Ltd. Address before: 510700 room 509, building D, 39 Ruihe Road, Huangpu District, Guangzhou City, Guangdong Province Patentee before: Guangzhou Shuimu Qinghua Technology Co.,Ltd. |