CN113054688B - Renewable energy cluster output distributed coordination control method considering uncertainty - Google Patents
Renewable energy cluster output distributed coordination control method considering uncertainty Download PDFInfo
- Publication number
- CN113054688B CN113054688B CN202110367986.6A CN202110367986A CN113054688B CN 113054688 B CN113054688 B CN 113054688B CN 202110367986 A CN202110367986 A CN 202110367986A CN 113054688 B CN113054688 B CN 113054688B
- Authority
- CN
- China
- Prior art keywords
- output
- power station
- cluster
- ord
- uncertainty
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000005457 optimization Methods 0.000 claims abstract description 59
- 238000009826 distribution Methods 0.000 claims abstract description 28
- 238000004088 simulation Methods 0.000 claims description 29
- 238000004422 calculation algorithm Methods 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 19
- 230000008569 process Effects 0.000 claims description 19
- 230000033228 biological regulation Effects 0.000 claims description 14
- 239000011159 matrix material Substances 0.000 claims description 8
- 230000001186 cumulative effect Effects 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 2
- 238000004891 communication Methods 0.000 abstract description 22
- 230000003993 interaction Effects 0.000 abstract description 7
- 230000007547 defect Effects 0.000 abstract description 5
- 239000006185 dispersion Substances 0.000 abstract description 4
- 230000000694 effects Effects 0.000 abstract description 4
- 238000000926 separation method Methods 0.000 abstract description 3
- 230000002411 adverse Effects 0.000 abstract description 2
- 238000004364 calculation method Methods 0.000 description 12
- 238000004458 analytical method Methods 0.000 description 5
- 238000012795 verification Methods 0.000 description 4
- 238000007476 Maximum Likelihood Methods 0.000 description 3
- 238000011217 control strategy Methods 0.000 description 3
- 238000007796 conventional method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000006872 improvement Effects 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 150000001875 compounds Chemical class 0.000 description 2
- 230000006735 deficit Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000035699 permeability Effects 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 230000001105 regulatory effect Effects 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000010835 comparative analysis Methods 0.000 description 1
- 230000001276 controlling effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 230000005611 electricity Effects 0.000 description 1
- 238000005538 encapsulation Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 230000000670 limiting effect Effects 0.000 description 1
- 230000004807 localization Effects 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 230000002688 persistence Effects 0.000 description 1
- 238000013439 planning Methods 0.000 description 1
- 238000010248 power generation Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000005096 rolling process Methods 0.000 description 1
- 239000000523 sample Substances 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012384 transportation and delivery Methods 0.000 description 1
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
- H02J3/466—Scheduling the operation of the generators, e.g. connecting or disconnecting generators to meet a given demand
-
- 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]
Landscapes
- Engineering & Computer Science (AREA)
- Power Engineering (AREA)
- Supply And Distribution Of Alternating Current (AREA)
Abstract
The invention discloses a renewable energy cluster output distributed coordination control method considering uncertainty, and belongs to the field of optimization control of power systems. In order to deal with the adverse effect of wind and light energy source uncertainty on control precision, aiming at the problem that the uncertainty of a traditional cluster output optimization model cannot be fully considered, an optimization model and a strategy considering the output uncertainty are constructed on the basis of TSL distribution; meanwhile, in the aspect of information architecture and implementation mode depending on coordination control, aiming at the defects of high computation complexity, long data acquisition period, high communication pressure and the like of the traditional centralized mode, in order to better adapt to the characteristics of frequent interaction of the provided optimization model data considering uncertainty, high communication timeliness requirement and the like, the distributed computation solution is realized based on the multi-agent system theory so as to reduce the computation scale and the communication pressure and realize fault separation and risk dispersion. Generally, the method and the device can improve the control precision, robustness and flexibility of the renewable energy cluster.
Description
Technical Field
The invention belongs to the field of optimization control of power systems, and particularly relates to a renewable energy cluster output distributed coordination control method considering uncertainty.
Background
In order to deal with the problems of energy and climate, the installed capacity of the wind, light and other renewable energy sources for power generation is rapidly increased. In the face of the emerging difficulty of large-scale bases in the aspects of consumption and delivery, the renewable energy development mode is gradually changed from a large-scale, long-distance and centralized traditional mode to a new mode of simultaneous localization and decentralization and local adjustment. The existing regulation and control mode usually does not limit the wind and light power supply when the power grid is sufficiently reserved, and the wind and light power supply is cut off in a large amount when the power grid is insufficiently reserved, so that the management is relatively extensive, and the phenomena of wind abandon and light abandon are obvious. Along with the continuous improvement of the permeability of the distributed renewable energy, the contradiction between the characteristics of multiple points and wide quantity of the power station and the insufficient adaptability of the existing regulation and control mode to the rapid expansion of the quantity of the power station is gradually highlighted, the full play of the consumption capability of the renewable energy is restricted, and the requirement of further optimization of the energy structure is difficult to meet. In the face of the challenges, a renewable energy clustering regulation and control method is formulated based on the ideas of integral encapsulation and hierarchical coordination, so that the observable, controllable and autonomous coordination of the whole cluster is realized, and the method has great significance for improving the accuracy and intelligent level of power grid regulation and control and relieving the pressure of dispatching personnel.
At present, the research and application aiming at the problem of cluster output control of wind and light renewable energy power stations mainly aim at a large-scale and centralized development base scene, and a multi-level centralized control system of a unit-power station-cluster is constructed. Under the framework, a renewable energy cluster is used as a whole to participate in output planning, scheduling and controlling of a superior control center, the cluster completes decomposition of instructions among power stations based on an output coordination optimization model after receiving superior output instructions, and the existing coordination optimization model is usually based on the day-ahead output prediction data of a wind power station or a photovoltaic power station. Specifically, output prediction data collected by the power station side is collected to a centralized control center arranged in a centralized control system, centralized calculation is carried out in the centralized control center, an optimization model is solved, and the solved output instruction is sent to the power station.
In fact, the current coordination optimization model cannot fully consider the influence of uncertainty of real-time output of renewable energy sources on regulation and control. With the rapid increase of the permeability of renewable energy sources, the output error caused by uncertainty of the renewable energy sources causes significant challenges to a power grid, and further influences the sufficient consumption of the renewable energy sources such as wind, light and the like. In addition, the existing centralized control mode depends on data interaction and centralized processing of a centralized control center, has high calculation complexity, long data acquisition period and high communication pressure, and is very sensitive to single-point faults of a communication network and the accuracy of data acquisition. In addition, the problems of difficult adaptation to wind, dispersion and differentiation of photoelectric stations and frequent access of new units exist, and the flexibility of system expansion is insufficient. With the rise of the distributed network structure, the power system gradually shows the development trend of local autonomous operation and intra-domain coordinated operation of a source and a load. However, the distributed control system has not been effectively applied to the output control problem of renewable energy clusters.
Disclosure of Invention
Aiming at the defects or improvement requirements of the prior art, the invention provides a renewable energy cluster output distributed coordination control method considering uncertainty, and aims to construct an output optimization model aiming at the problem of control errors caused by uncertainty of output of renewable energy so as to improve the control precision of a renewable energy cluster, and simultaneously design a distributed algorithm aiming at the defect of centralized optimization control, realize decentralized coordination of output of each power station of the cluster, and improve the robustness and flexibility of cluster control.
In order to achieve the above object, the present invention provides a distributed coordination control method for renewable energy cluster output considering uncertainty, including:
s1, simulating output uncertainty errors of a renewable energy cluster by using a TLS (transport layer security) distribution model;
s2, constructing an output optimization model of each power station of the cluster by taking minimum electric power abandoned by the wind and photoelectric stations as an optimization target; the output optimization model is used for correcting the output instruction of the power station according to the uncertainty error simulation result in a set time period;
s3, solving the optimization model based on a multi-agent system and a consistency principle thereof to obtain output adjustment values of each power station;
and S4, changing the active power output of the unit or starting and stopping the unit according to the output adjustment value of each power station, thereby performing active power control.
Further, S1 includes:
01. modeling the distribution of the power station output prediction error;
calculating TLS distribution parameters of the prediction errors based on historical statistical data of the power station output prediction errors and probability density functions of the prediction errors, and thus obtaining TLS distribution curves of the power station output prediction errors; wherein the probability density function of the prediction error is:
f (x) and F (x) are the probability density function and cumulative distribution function, P, of TLS, respectively cap The installed capacity of the power station;
02. simulating uncertainty errors of the power station output;
randomly generating error simulation data from the distribution curve, and simulating P for the ith simulation err，(i) When P is err，(i) ∈[-P pre，(i) ,P cap -P pre，(i) ]If not, refusing to carry out the next simulation;
and (4) simulating each power station for multiple times until the power station receives effective simulation for a set number of times, and taking the average value of the simulation values of the times as the error simulation value of the power station at the moment.
Further, the objective function of the output optimization model is:
in the formula, P pre,i (t) is a predicted value of the day-ahead output of the power station i in the time period t; p ord,i (t) is a command value that the power station i should output power in the time period t; and N is the number of renewable energy power stations in the cluster domain.
Further, the constraints of the output optimization model include:
clustering active balance constraints:
P ord,clu (t) is the total output instruction value of the cluster at the t time period;
and power station real-time output constraint: p is more than or equal to 0 ord,i (t)≤P pre,i (t)≤P cap,i (t)，P cap,i (t) installed capacity of station i;
power station output regulation rate constraint: p ord,i (t)-P ord,i (t-Δt)≤|α i P cap,i 0，α i Is the regulation rate factor of the plant i.
Further, step S3 specifically includes,
setting the power station nodes as coordination agents, executing the following distributed iterative computation process, and when the consistent variables of all the coordination agents in the cluster are equal, converging the algorithm to obtain an optimal solution, so as to obtain an output adjustment value of the power station p in the control period:
01. updating the consistency variable lambda of the iteration node p of the k +1 round according to the state of the iteration of the k round p (k+1)；
n is the number of power stations contained in the cluster, and the matrix element a p,q Representing the weight of an edge between any two nodes p, q,
for consistent convergence factor, mu p (k) Is a force adjustment term;
02. calculating the output P of the node P of the k +1 wheel according to the consistency variable of the k +1 wheel ord,p (k+1)：
P ord,p (k+1)＝-[λ p (k+1)(P pre,p +P err,p ) 2 ]/2+P pre,p +P err,p
The following corrections were made to the output after each update cycle:
03. according to the updated P ord,p Calculating (k +1) to obtain an output adjustment term mu in k +1 iteration p Value of (k + 1): mu.s p (k+1)＝μ p (k)-[P ord,p (k+1)-P ord,p (k)]。
Further, when a cluster output error out-of-limit is detected, the method further comprises the steps of:
calculating the output compensation expected value delta P of each power station k (t)＝P pre,i (t)+P err,i (t)－P ord,i (t) selecting Δ P k (t) the power station with the largest value is used as a compensation power station i;
if the compensation power station i satisfies:
P pre,i (t)+P err,i (t)-P ord,i (t)＞P ord,clu (t)-P real,clu (t)
in the formula, P ord,clu (t) is the output command value of the cluster in the period t, P real,clu (t) is an actual output value of the cluster at the initial moment of the period t;
the output instruction of the compensation power station i in the next control period is corrected to
Clustering N-1 stations other than the compensating station i
Participating in optimization, and setting output reference values in a t +1 period according to an optimization result;
if the compensation power station i satisfies:
P pre,i (t)+P err,i (t)-P ord,i (t)＜P ord,clu (t)-P real,clu (t)
and correcting the output instruction of the compensation power station i in the next period as follows:
and according to the output compensation expected value delta P k (t) sequentially selecting the next compensation power station k from large to small, and correcting the output instruction of the power station k in the next control period into
Meanwhile, the N-2 power stations except the compensation power stations i and k participate in optimization in the cluster based on the following formula, and output reference values in a t +1 period are set according to an optimization result:
in general, the above technical solutions contemplated by the present invention can achieve the following advantageous effects compared to the prior art.
1) According to the invention, the uncertainty of the renewable energy power station output is subjected to probability modeling, and a cluster output optimization control strategy considering the uncertainty is constructed, so that the problem of control errors caused by the contradiction that the current prediction data is difficult to accurately match with the actual data can be solved, the cluster output can be enabled to more accurately track the regulation and control instruction of the upper-level dispatching center, the output fluctuation is effectively reduced, the friendly grid connection of renewable energy clusters is realized, a foundation is laid for the problems that the wind and light energy of cluster levels subsequently directly participate in system frequency modulation and the like, and the utilization depth of renewable energy is improved.
2) Aiming at the defects of centralized interactive computation in the traditional cluster operation, the distributed coordination algorithm is constructed based on the multi-agent system theory, so that the independent autonomy and the operation information privacy in a cluster area can be kept, the computation scale and the communication pressure can be reduced, the fault separation and risk dispersion are realized, the changeable communication topological structure can be effectively dealt with, the good robustness and flexibility are shown in different operation scenes, and the technical purposes of cluster friendly grid connection and self-adaptive coordination are finally realized.
3) The invention makes a supplement strategy aiming at extreme meteorological conditions, avoids cluster output continuous deviation instructions caused by strong uncertainty and strong fluctuation of output of individual power stations under certain extreme meteorological conditions, and further improves the control precision of the cluster output.
Drawings
Fig. 1 shows the general framework of the method in a multi-level organization-station-cluster relationship for the physical structure and information flow hierarchy.
Fig. 2 shows the solution flow of the distributed algorithm.
Fig. 3 shows the control principle of the station level active control system.
FIG. 4 shows a main wiring diagram of a system used for example verification.
In fig. 5, (a) is an iterative convergence process of the active power output of each power station; fig. 5(b) shows a process of tracking the scheduling instruction for the total cluster output.
Fig. 6 (a) shows the convergence process of the plant output; fig. 6 (b) shows a process of changing the cluster output imbalance.
Fig. 7 shows the real-time output of the cluster during operation, wherein 4 curves respectively show the power prediction before the day, the real-time regulation instruction, the output curve corresponding to the method of the present invention, and the output curve corresponding to the conventional method.
Fig. 8 shows the difference between the real-time output and the control command during the operation of the cluster.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention. In addition, the technical features involved in the embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.
In order to deal with the adverse effect of the inherent uncertainty of wind and light energy on the control precision, aiming at the problem that the uncertainty of the traditional cluster output optimization model cannot be fully considered, the invention constructs an optimization model and a strategy considering the output uncertainty based on the TSL distribution (T distribution-scale distribution, namely T distribution containing scale parameters and position parameters). Meanwhile, in the aspect of information architecture and implementation mode depending on coordination control, aiming at the defects of high computation complexity, long data acquisition period, high communication pressure and the like of the traditional centralized mode, in order to better adapt to the characteristics of frequent data interaction, high communication timeliness requirement and the like of the provided optimization model considering uncertainty, the distributed computation solution is realized based on the multi-agent system theory so as to reduce the computation scale and the communication pressure and realize fault separation and risk dispersion. Generally, the method and the device can improve the control precision, robustness and flexibility of the renewable energy cluster.
As shown in fig. 1, the present invention proposes a renewable energy cluster output distributed control method considering uncertainty. Fig. 1 constructs a typical renewable energy cluster scene containing wind power plants and photovoltaic power plants in a multi-level relation of unit-power plant-cluster, and gives an overall framework of the method in the cluster example from the aspects of physical structure and information flow direction. In the multi-agent system theory, a cluster is composed of a group of agents which are closely related in physical and information levels, and each agent has initiative and self-adaptability. In the example, each power station is modeled as a coordination agent, and a hub agent is added in a cluster grid-connected point (only one hub agent needs to be arranged in a cluster domain), so that a multi-agent system is constructed. The junction intelligent agent has three functional units, namely instruction receiving, grid-connected point measurement and power station information acquisition, and does not need to undertake calculation tasks. And the superior dispatching center (provincial dispatching) acquires the cluster running state information in real time, makes an output plan instruction for the whole cluster and receives the output plan instruction by the junction intelligent agent. The coordination agent performs the local distributed computing function of the power station, and the decentralized solution of the optimization model can be realized only by carrying out information interaction with the adjacent power station and carrying out local computation. Fig. 2 shows the detailed flow of the distributed algorithm in a complete cycle from the time scale level.
In a complete embodiment, the method comprises the following steps:
and (1) analyzing the output uncertainty of the renewable energy power station level according to the T location-scale distribution to form a power station output uncertainty error simulation method.
Related research has been directed to accounting for the uncertainty of renewable energy sources in the regulation of power systems containing renewable energy sources, but often analysis is based on normal distributions. In fact, many leading-edge studies indicate that the plant-level uncertainty error of the wind-solar renewable energy source does not follow a normal distribution, but is related to the magnitude of the predicted output, and shows a persistence in the time sequence characteristic, and the TLS (T-location-scale) distribution, i.e., the T-distribution containing the scale parameter and the position parameter, has a better fitting effect. The method comprises the following steps:
(1.1) modeling of uncertainty error probability distribution
Estimating the distribution of the plant output prediction error, the prediction error can be considered approximately to obey the following TLS probability distribution:
P err ～TLS(k 1 P pre ,k 2 P pre +k 3 ,v) (1)
in the formula, k is more than or equal to-1 1 ≤0,k 2 ＞0,k 3 ＞0,v＞0，P pre Is a predicted value of the day-ahead output of the power station, P err The error is predicted for the force. The probability density function f (x) of the generic TLS distribution can be written as:
wherein Γ (x) is a gamma function, μ is a position parameter, σ is a scale parameter, and ν is a shape parameter, respectively corresponding to k of formula (1) 1 P pre 、k 2 P pre +k 3 And v correspond to each other.
In the application scene of the invention, the output of the renewable energy power station has a value range limit value, the output of each power station is more than 0 and less than the installed capacity, namely P err +P pre ∈[0,P cap ]. And formula (2) will cover [ - ∞, + ∞]Any value of the interval, which is not in accordance with the physical background of the present invention, needs to be modified to ensure that the function falls within 0, P cap ]The probability in the value domain is 1. The probability density function of the prediction error can be rewritten as:
wherein f (x) and F (x) are the probability density function and cumulative distribution function, respectively, of TLS, P cap For installations of power stationsThe machine capacity is also the maximum value of the power station output.
To obtain a specific distribution curve, the characteristic parameters of the TLS distribution need to be solved.
Computing using maximum likelihood estimation except for the predicted output P pre Four unknown parameters. Obtaining the logarithm of the maximum likelihood function based on the statistical error data and the probability density function:
wherein M is the total number of data samples. When the logarithm of the maximum likelihood function takes the maximum value, the values of the four parameters are optimal, and the probability distribution is closest to the actual situation. ln L can not write out specific function expression to solve the analytic solution, and can adopt optimization algorithms such as particle swarm algorithm and the like to directly solve the numerical solution.
(1.2) uncertainty error simulation method
The uncertainty error of the power station output is simulated, and the method comprises the following processes:
and (3) calculating historical statistical data of prediction errors of a certain power station at a certain moment according to the method (1.1), and obtaining characteristic parameters according to the formula (4) so as to fit and obtain a TLS distribution curve of the power station. Then, for each simulation, error simulation data is randomly generated from the distribution, and P is simulated for the ith simulation err，(i) When P is err，(i) ∈[-P pre，(i) ,P cap -P pre，(i) ]If the simulation value is not accepted, the simulation value is rejected and the next simulation is performed. And (4) simulating each power station for multiple times until the power station receives effective simulation for K times, and taking the average value of the simulation values of the times as the error simulation value of the power station at the moment.
And (2) constructing a coordination control strategy of the cluster output based on an uncertainty error simulation method.
The power system usually takes 15min as a dispatching cycle, and for adapting to the dispatching cycle, the coordination control strategy takes a delta t which is not more than the dispatching cycle and is in integral multiple relation 1 To control the period (in)In this example,. DELTA.t 1 Taking for 5 min). In actual operation, errors in the plant day-ahead prediction will cause the total cluster output to deviate from the command. If the output prediction of partial power stations has large negative errors, the actual output capacity is smaller than the output instruction, so that the output of the power station is insufficient, and the power stations with surplus generating capacity still track the own instruction and cannot be effectively matched to cause extra wind and light abandon. In summary, the following strategies are made:
firstly, the power station side calculates according to the accumulated historical data of the power station to obtain an uncertainty error analog value P err,i (t) of (d). The minimum electric power abandon of the wind and photoelectric stations is taken as an optimization target, and uncertainty error P is considered err,i (t), defining the coordination optimization target of the output of each power station of the cluster in the t period as follows:
in the formula, P pre,i (t) the predicted value of the day-ahead output of the power station i in the time period t is a known constant; p is ord,i (t) is the instruction value that the power station i should output in the time period t, and is N decision variables of the optimization problem; and N is the number of renewable energy power stations in the cluster domain. The constraints of the optimization function are as follows:
1. cluster active balance constraints
In the formula, P ord,clu And (t) is the total output instruction value of the cluster t time period.
2. Real-time output constraint of power station
0≤P ord,i (t)≤P pre,i (t)≤P cap,i (t) (7)
In the formula, P cap,i (t) is the installed capacity of the station i.
3. Power plant output regulation rate constraints
P ord,i (t)-P ord,i (t-Δt)≤|α i P cap,i | (8)
In the formula, alpha i And the adjusting rate coefficient of the power station i is determined by the actual performances of the wind power plant and the photovoltaic power station.
Particularly, under some extreme meteorological conditions, when the optimization model is adopted, because the output of an individual power station presents strong uncertainty and strong fluctuation, the cluster output may still deviate from the instruction continuously, and the following supplementary strategies are made for the extreme conditions:
the out-of-limit coefficient β is defined, and an allowable out-of-limit range is 5% if β ═ 1.05. If the time is in the control period t, the measuring unit of the hub intelligent agent detects that the cluster output error is out of limit, and calculates the output compensation expected value delta P of each power station k (t)＝P pre,i (t)+P err,i (t)－P ord,i (t) selecting Δ P k The station with the largest value of (t) is used as the compensation station i. If the power station has
P pre,i (t)+P err,i (t)-P ord,i (t)＞P ord,clu (t)-P real,clu (t) (9)
In the formula, P ord,clu (t) is the output command value, P, of the cluster in the period t real,clu And (t) is the actual output value of the cluster at the initial moment of the period t. If equation (9) is satisfied, the output reference value of the plant i in the next control period (t +1) is set as
Is represented as follows:
meanwhile, N-1 power stations of the cluster except the power station i participate in optimization based on the formula (11), and output reference values in a t +1 period are set according to an optimization result and are expressed as follows:
the active balance constraint of equation (11) is modified as:
if formula (9) does not hold, i.e.
P pre,i (t)+P err,i (t)-P ord,i (t)＜P ord,clu (t)-P real,clu (t) (13)
The output instruction of the compensation power station i in the next period is corrected to
Then, the desired value Δ P is compensated for according to the output k (t) sequentially selecting the next compensation power station k from large to small, and correcting the output instruction of the power station k in the next control period into
Is represented as follows:
meanwhile, N-2 power stations of the cluster except the power stations i and k participate in optimization based on the formula (16), and output reference values in t +1 periods are set according to an optimization result and are represented as follows:
the active balance constraint of equation (16) is modified as:
and (3) constructing a distributed coordination algorithm, solving the output optimization model, and realizing rolling optimization control. The theory and consistency principle of the multi-agent system can be effectively applied to a cluster distributed computing scene containing a large number of nodes, so that a coordination algorithm is constructed based on the theory to solve the optimization problem in the formula (5), and coordination and junction agents in the cluster multi-agent system are shown in fig. 1. The algorithm is constructed as follows:
(3.1) extracting consistency variables
The multi-agent system represents the state of the agent by defining the consistency variable, and then iteratively calculates the consistency variable to solve the optimization problem, so that the consistency variable capable of reflecting the key state of the agent node is extracted according to the optimization model. In the step (2), the optimization target is as follows:
by applying a classical Lagrange multiplier method, let lambda represent the Lagrange multiplier corresponding to the cluster power balance equality constraint, and when the inequality constraint is not considered, the original multi-objective optimization problem can be converted into:
in the formula (I), the compound is shown in the specification,
indicating the difference in active balance after the cluster receives the new output command. In this optimization problem, the output P of N power stations within a cluster ord,i (t) is N decision variables, P pre,i (t)、P ord,clu (t)、P err,i (t) are all known amounts.
Applying the first-order optimality condition of the power station i to the decision quantity and the Lagrange multiplier and solving the partial derivative to obtain the optimality condition of the equivalent unconstrained optimization problem:
the optimality condition is a distributed calculation coordination equation, the equation can be obtained, and when the algorithm achieves the optimal convergence:
defining a consistency variable lambda of the intelligent agent node power station i under the condition of not considering equation constraint based on the optimal convergence equivalent condition given by the formula (21) i Comprises the following steps:
thus, consistent variable extraction of the optimization problem is completed. Lambda of all agent nodes i When the values are equal in iterative calculation, the algorithm converges to obtain an optimal solution.
(3.2) constructing an adjacency matrix
After the consistency variables are defined, an adjacency matrix is established according to the communication interaction relationship among the intelligent agent nodes of the cluster. Let G denote the communication topology of each power station in the cluster. Each plant represents a node, i.e. the system contains n nodes.
If a communication interaction channel exists between two nodes, the two nodes are called to be communicated (a channel existing between the nodes is called as an edge). The topology of the graph can be formed by a symmetrical n multiplied by n adjacent matrix A ═ a p,q ) n×n Wherein the weight of an edge between any two nodes p, q is represented by a matrix element a p,q And (4) showing. Defining the elements a of the adjacency matrix A p,q Comprises the following steps:
in the formula D p Is a set of nodes with which a communication path exists, the set including the node p itself. The sum of the number of nodes having a path with the node p is d p Degree called node p, d p ＝|D p L. To ensure algorithm convergence, at least one path must exist between any node in the graph and other nodes.
(3.3) distributed iterative computation flow
A complete round of iterative computation is divided into the following three steps:
1) updating lambda of k +1 iteration according to state of k iteration p (k+1)
Let x p In the iterative calculation process, each node and the adjacent node with communication relation carry out information interaction, and then the consistency variable value in the current round is obtained by the iterative update of the previous round of state information of the node and the adjacent node, wherein the process is expressed as follows:
in the formula, k is the time step number, i.e., the iteration step number of the iterative computation.
Based on the row random property of the adjacent square matrix a, equation (24) is further simplified, and the following results are obtained:
specifically, in terms of the optimization problem, for a power station p, if there are n power stations that can communicate with p, the update equation of the consistency variables of the power station p can be expressed as:
at λ i In the modeling process, the constraint of a power balance equation is not considered, and a consistency convergence coefficient is introduced
And the output adjustment term mu p (k) For satisfying the power balance, equation (26) is modified to:
in the formula (I), the compound is shown in the specification,
the consistency convergence coefficient is a positive scalar quantity, and influences the convergence speed of consistency iterative computation; mu.s p (k) A force adjustment term for node p, a probe value, mu, representing the force deficit of the node pair cluster p (k) See step 3 of iteration.
2) Computing P from k +1 round consistency variables ord,p
According to the formula (22), the output of the k +1 wheel node p is updated to
P ord,p (k+1)＝-[λ p (k+1)(P pre,p +P err,p ) 2 ]/2+P pre,p +P err,p (28)
After each round of updating, whether the threshold is out of limit or not needs to be judged, and the out-of-limit condition is corrected as follows:
3) updating the output adjustment term mu p (k) And feed back
According to the updated P ord,p Calculating (k +1) to obtain an output adjustment term mu in k +1 iteration p The value of (k +1), the equation, is as follows:
μ p (k+1)＝μ p (k)-[P ord,p (k+1)-P ord,p (k)] (30)
and when the consistency variables of all the coordination agents in the cluster are equal, the algorithm converges to obtain an optimal solution, namely the output regulating value of the power station p in the control period.
(3.4) response of output command between station and unit
After the power station side obtains the output regulating value of the station through distributed calculation, the power stationAnd the AGC system changes the active power output of the unit or starts and stops the unit through communication with the unit, so that active control is performed. Fig. 3 shows a typical plant-to-plant control framework. In fig. 3, the power station output control system uses a proportional-integral control algorithm to obtain the area adjustment power, and then distributes the adjustment power to each unit. In the figure, K p And K i Proportional coefficient and integral coefficient, respectively, and beta is the frequency deviation coefficient of the control area.
The steps 1 to 3 are the complete process of the method, and the method is subjected to example verification analysis based on a renewable energy cluster scene constructed by a CEPRI example system, wherein the main wiring of the example system is shown in FIG. 4. In this system, clusters of 7 power stations are shown within the dotted line. The total installed capacity of the cluster is 439MW, wherein the power stations 1, 2, 3, 5 and 6 are double-fed induction unit wind power plants, and the nodes 4 and 7 are centralized inverter photovoltaic power stations. The sampling step length Δ k is 0.01s (representing communication delay 10ms), the convergence threshold ξ is 0.1MW, and the consistency convergence coefficient
。
The initial value of the output force of the cluster is set to 233MW, and then the cluster receives a plurality of output force adjusting instructions. The real-time optimization model was solved by the consistency algorithm, and the results are shown in table 1 and fig. 5. In fig. 5, (a) is an iterative convergence process of the active power output of each power station; fig. 5(b) shows a process of tracking the scheduling command for the total cluster contribution.
TABLE 1 Algorithm solution results
The convergence coefficient is a main parameter influencing the convergence speed of consistency iteration and is matched with the power adjustment item to adjust the power balance. Convergence of the algorithm is discussed by setting convergence coefficients with different sizes, the influence of the difference of cluster sizes on the convergence is considered, and 14 power station nodes and 21 power station nodes are respectively constructed according to the equal-proportion expansion of the installed scale of the existing clusterThe system was verified and the results are shown in table 2. As can be seen from table 2, it is,
if the value is larger, the number of iteration steps is less, and the algorithm convergence speed is higher; the larger the cluster size, the more iteration steps are required and the longer the convergence time is required. In addition to this, the present invention is,
is not greater, the better, the greater
The change amount of each iteration of the consistent variable is correspondingly enlarged, so that the output of the power station is not smoothly close to the optimal solution any more, the oscillation performance is increased in the iteration process, and finally the non-convergence can be caused. Besides the convergence of the method, the comparative analysis with the centralized computing method is also necessary to verify the application value of the invention.
TABLE 2 Convergence analysis results
Solving the same problems by respectively adopting a Genetic Algorithm (GA) and a Cplex12.5 solver (interior point method) as a centralized computing method, wherein the instruction takes a larger value and the parameters
The results are shown in Table 3. From the result, under the condition that the convergence threshold value ξ is 0.1MW, each calculation method can converge, but the distributed consistency algorithm has certain advantages in the calculation speed along with the enlargement of the cluster size.
TABLE 3 centralized versus distributed comparison
In order to verify the robustness of the algorithm, the simulation scene is mainly modified by two points: 1) a plurality of communication channels in the cluster are failed, so that the communication topology of the system is changed, but the communication channels are still connected; 2) a 'plug and play' scene is simulated in the operation process, namely, a power station is quitted or added in the midway in the operation process. Still setting the cluster initial power to 233MW, and the cluster initial scheduling instruction to 310 MW; the power station 7 quits operation due to faults at the 8 th time, and is connected to the power grid again at the 18 th time; the simulation results are shown in fig. 6. Fig. 6 (a) shows the convergence process of the plant output; fig. 6 (b) shows a process of changing the cluster output imbalance. The analysis result shows that, on one hand, although the communication topology is changed and the communication condition becomes worse, the algorithm can still effectively and rapidly converge, and the iteration number required by convergence is not obviously increased; on the other hand, the distributed consistency algorithm can effectively adapt to the situation of 'plug and play' of the dynamic exit or addition of the intelligent agent, and has good real-time operation robustness.
In order to test the effectiveness of the method in practical application, simulation verification is carried out according to actual data of a certain renewable energy cluster in northwest China at a day of 11: 00-15: 00. The time interval is divided into 48 time points with the resolution of 5min, wherein the phenomenon of obviously limiting electricity, abandoning wind and abandoning light occurs in 12: 20-13: 25 of the day. The calculation is performed by using the method model provided herein and the conventional method without considering uncertainty, respectively, as a comparison verification, and the results are shown in fig. 7 and fig. 8. Fig. 7 shows the real-time output of the cluster during operation, wherein 4 curves respectively show the power prediction before the day, the real-time regulation instruction, the output curve corresponding to the method of the present invention, and the output curve corresponding to the conventional method. Fig. 8 shows the difference between the real-time output and the control command during the operation of the cluster. Quantifying the deficit of the output control, and defining the average relative error delta as:
calculated as 11: 00-15: the average relative error δ for the two strategies was 0.37% and 1.78% for the total output of the cluster over time 00. Analysis shows that after the method is applied, because the influence of inherent errors of renewable energy source prediction in the day-ahead on real-time output of each power station is weakened, the output error in the integral operation process of the cluster is obviously smaller than that of a traditional model without considering uncertainty, and power fluctuation caused by output uncertainty of the power station can be better suppressed.
It will be understood by those skilled in the art that the foregoing is only a preferred embodiment of the present invention, and is not intended to limit the invention, and that any modification, equivalent replacement, or improvement made within the spirit and principle of the present invention should be included in the scope of the present invention.
Claims (5)
1. A renewable energy cluster output distributed coordination control method considering uncertainty is characterized by comprising the following steps:
s1, simulating output uncertainty errors of a renewable energy cluster by using a TLS (transport layer security) distribution model; s1 includes:
01. modeling the distribution of the power station output prediction error;
calculating TLS distribution parameters of the prediction errors based on historical statistical data of the power station output prediction errors and probability density functions of the prediction errors, and thus obtaining TLS distribution curves of the power station output prediction errors; wherein the probability density function of the prediction error is:
f (x) and F (x) are the probability density function and cumulative distribution function, P, of TLS, respectively cap The installed capacity of the power station; p err Indicating the predicted error of the output, P pre Representing a predicted value of the day-ahead output of the power station;
02. simulating uncertainty errors of power station output;
randomly generating error simulation data from the distribution curve, and simulating P for the ith simulation err，(i) When P is err，(i) ∈[-P pre，(i) ,P cap -P pre，(i) ]If not, refusing to carry out the next simulation;
simulating each power station for many times until the effective simulation of the power station reaches the set times, and taking the average value of the simulation values of each time as the error simulation value of the power station at the moment;
s2, constructing an output optimization model of each power station of the cluster by taking minimum electric power abandoned by the wind and photoelectric stations as an optimization target; the output optimization model is used for correcting the output instruction of the power station according to the uncertainty error simulation result in a set time period; the objective function of the output optimization model is as follows:
in the formula, P pre,i (t) is a predicted value of the day-ahead output of the power station i in the time period t; p ord,i (t) is a command value that the power station i should output power in the time period t; n is the number of renewable energy power stations in the cluster domain;
s3, solving the optimization model based on a multi-agent system and a consistency principle thereof to obtain output adjustment values of each power station;
and S4, changing the active power output of the unit or starting and stopping the unit according to the output adjustment value of each power station, thereby performing active power control.
2. The distributed coordination control method for renewable energy cluster output according to claim 1, wherein the constraints of said output optimization model include:
cluster active balance constraint:
P ord,clu (t) is the total output instruction value of the cluster at time period t;
and power station real-time output constraint: p is more than or equal to 0 ord,i (t)≤P pre,i (t)≤P cap,i (t)，P cap,i (t) installed capacity of station i;
power station output regulation rate constraint: p ord,i (t)-P ord,i (t-△t)≤|α i P cap,i |，α i Is the regulation rate factor of the plant i.
3. The distributed coordination control method for renewable energy cluster output based on uncertainty as claimed in claim 1, wherein step S3 specifically comprises,
setting the power station nodes as coordination agents, executing the following distributed iterative computation process, and when the consistent variables of all the coordination agents in the cluster are equal, converging the algorithm to obtain an optimal solution, so as to obtain an output adjustment value of the power station p in the control period:
01. updating the consistency variable lambda of the iteration node p of the k +1 round according to the state of the iteration of the k round p (k+1)；
n is the number of power stations contained in the cluster, and the matrix element a p,q Representing the weight of an edge between any two nodes p, q,
for consistent convergence factor, mu p (k) Is a force adjustment term;
02. calculating the output P of the node P of the k +1 wheel according to the consistency variable of the k +1 wheel ord,p (k+1)：
P ord,p (k+1)＝-[λ p (k+1)(P pre,p +P err,p ) 2 ]/2+P pre,p +P err,p The following corrections were made to the output after each update cycle:
03. according to the updated P ord,p Calculating (k +1) to obtain an output adjustment term mu in k +1 iteration p Value of (k + 1): mu.s p (k+1)＝μ p (k)-[P ord,p (k+1)-P ord,p (k)]。
4. The distributed coordination control method for renewable energy cluster output based on uncertainty according to any of claims 1-3, wherein when cluster output error out-of-limit is detected, said method further comprises the following steps:
calculating the output compensation expected value delta P of each power station k (t)＝P pre,i (t)+P err,i (t)－P ord,i (t) selecting DeltaP k (t) the power station with the largest value is used as a compensation power station i; p pre,i (t) is a predicted value of the day-ahead output of the station i during the time period t, P ord,i (t) a command value indicating that the plant i should output force during the period t;
if the compensation power station i satisfies:
P pre,i (t)+P err,i (t)-P ord,i (t)>P ord,clu (t)-P real,clu (t)
in the formula, P ord,clu (t) is the output command value of the cluster in the period t, P real,clu (t) is an actual output value of the cluster at the initial moment of the period t;
the output instruction of the compensation power station i in the next control period is corrected to
Clustering N-1 plants except the compensation plant i is based on the formula:
participating in optimization, and setting output reference values in a t +1 period according to an optimization result;
if the compensation power station i satisfies:
P pre,i (t)+P err,i (t)-P ord,i (t)<P ord,clu (t)-P real,clu (t)
and correcting the output instruction of the compensation power station i in the next period as follows:
and compensating the expected value DeltaP according to the output k (t) sequentially selecting the next compensation power station k from large to small, and correcting the output instruction of the power station k in the next control period into
Meanwhile, N-2 power stations except the compensation power stations i and k participate in optimization according to the following formula, and output reference values in a t +1 period are set according to an optimization result:
5. a renewable energy clustered output distributed coordinated control system that accounts for uncertainty, comprising: a computer-readable storage medium and a processor;
the computer-readable storage medium is used for storing executable instructions; the processor is configured to read executable instructions stored in the computer-readable storage medium, and execute the renewable energy cluster output distributed coordination control method considering uncertainty according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110367986.6A CN113054688B (en) | 2021-04-06 | 2021-04-06 | Renewable energy cluster output distributed coordination control method considering uncertainty |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110367986.6A CN113054688B (en) | 2021-04-06 | 2021-04-06 | Renewable energy cluster output distributed coordination control method considering uncertainty |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113054688A CN113054688A (en) | 2021-06-29 |
| CN113054688B true CN113054688B (en) | 2022-09-16 |
Family
ID=76517558
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110367986.6A Expired - Fee Related CN113054688B (en) | 2021-04-06 | 2021-04-06 | Renewable energy cluster output distributed coordination control method considering uncertainty |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113054688B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113659631B (en) * | 2021-08-09 | 2022-07-26 | 大连理工大学 | Wind-solar power station group output description method considering time-varying characteristics |
| CN114859761B (en) * | 2022-04-12 | 2023-05-23 | 华能澜沧江水电股份有限公司 | Intelligent control method and system for interactive power station |
| CN115642650B (en) * | 2022-12-26 | 2023-05-16 | 中国华能集团清洁能源技术研究院有限公司 | Method and system for determining micro-grid operation strategy in isolated grid mode |
| CN116646973B (en) * | 2023-04-28 | 2024-05-03 | 国网山东省电力公司日照供电公司 | Information interaction method of micro-grid, micro-grid and storage medium |
Family Cites Families (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| IT201600131878A1 (en) * | 2016-12-28 | 2018-06-28 | Electro Power Systems Mfg S R L | MICRORETAL CONTROL SYSTEM FOR THE PRODUCTION AND DISTRIBUTION OF ELECTRIC ENERGY FROM DIFFERENT SOURCES OF PRODUCTION, AND ITS CONTROL METHOD |
| CN106877398B (en) * | 2017-03-23 | 2020-05-29 | 燕山大学 | Micro-power source distributed coordination control method based on multiple intelligent agents |
| CN108054790B (en) * | 2017-12-27 | 2021-03-16 | 云南电网有限责任公司 | Wind-solar power generation cluster active real-time optimization control method based on predicted output successive approximation |
| CN109560574A (en) * | 2018-12-14 | 2019-04-02 | 国网江苏省电力有限公司苏州供电分公司 | A kind of intelligent distribution network space truss project method considering uncertain factor |
| CN112165084B (en) * | 2020-07-24 | 2022-02-15 | 国网内蒙古东部电力有限公司通辽供电公司 | Multi-time scale optimization method considering photovoltaic-load bilateral prediction uncertainty |
-
2021
- 2021-04-06 CN CN202110367986.6A patent/CN113054688B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN113054688A (en) | 2021-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN113054688B (en) | Renewable energy cluster output distributed coordination control method considering uncertainty | |
| Li et al. | A distributed coordination control based on finite-time consensus algorithm for a cluster of DC microgrids | |
| CN107769254B (en) | A kind of wind-powered electricity generation cluster trajectory predictions and hierarchical control method | |
| CN107039977B (en) | Robust scheduling uncertainty set construction method for power system | |
| CN108376989B (en) | Battery energy storage power station partition control method and system based on multiple intelligent agents | |
| CN106875033B (en) | Wind power cluster power prediction method based on dynamic self-adaption | |
| CN110826880B (en) | Active power distribution network optimal scheduling method for large-scale electric automobile access | |
| CN108512258B (en) | Wind power plant active scheduling method based on improved multi-agent consistency algorithm | |
| CN108494022B (en) | Accurate scheduling method based on distributed power supply in micro-grid | |
| CN108695857B (en) | Automatic voltage control method, device and system for wind power plant | |
| CN111756050B (en) | Distributed voltage coordination control method and system for wind power plant group | |
| CN105207272B (en) | The random economic load dispatching method and device of Electrical Power System Dynamic based on general distribution | |
| CN106026084B (en) | A kind of AGC power dynamic allocation methods based on virtual power generation clan | |
| CN110137981B (en) | Distributed energy storage aggregator AGC method based on consistency algorithm | |
| CN109378863B (en) | Power grid real-time power generation control optimization decision method based on weight and constraint correlation adjustment | |
| CN107069835B (en) | Real-time active distribution method and device for new energy power station | |
| CN110676849A (en) | Method for constructing islanding micro-grid group energy scheduling model | |
| CN111860950A (en) | Probability assessment method for power adjustability of virtual power plant | |
| CN115632416A (en) | Energy storage system distributed control strategy optimization method and device | |
| CN115795992A (en) | Park energy Internet online scheduling method based on virtual deduction of operation situation | |
| CN105720574B (en) | The data drive control method of electric system list region LOAD FREQUENCY based on SPSA | |
| CN113298376B (en) | Given time consistency control method for economic dispatching of smart power grid with valve point effect | |
| CN113837449B (en) | Centralized optimization scheduling method for power grid system participated by virtual power plant | |
| CN115021327A (en) | Collaborative day-ahead optimization scheduling method of interconnected micro-grid system based on virtual energy storage | |
| Dapoz et al. | Mixed regression clustering techniques for the medium term prediction of reactive power within transmission grids |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20220916 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |