CN116843066A - Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant - Google Patents

Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant Download PDF

Info

Publication number
CN116843066A
CN116843066A CN202310784504.6A CN202310784504A CN116843066A CN 116843066 A CN116843066 A CN 116843066A CN 202310784504 A CN202310784504 A CN 202310784504A CN 116843066 A CN116843066 A CN 116843066A
Authority
CN
China
Prior art keywords
power
data
renewable energy
cluster
output
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.)
Pending
Application number
CN202310784504.6A
Other languages
Chinese (zh)
Inventor
张雁茹
朱建军
许万源
戴璟
祁晓乐
王振江
闫爱云
金硕巍
王长瑞
张沐瑶
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NATIONAL BIO ENERGY GROUP CO LTD
Tsinghua University
Original Assignee
NATIONAL BIO ENERGY GROUP CO LTD
Tsinghua University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by NATIONAL BIO ENERGY GROUP CO LTD, Tsinghua University filed Critical NATIONAL BIO ENERGY GROUP CO LTD
Priority to CN202310784504.6A priority Critical patent/CN116843066A/en
Publication of CN116843066A publication Critical patent/CN116843066A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G06F18/232—Non-hierarchical techniques
    • G06F18/2321—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23213—Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with fixed number of clusters, e.g. K-means clustering
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/24—Classification techniques
    • G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
    • G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q30/00—Commerce
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0283—Price estimation or determination
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Energy or water supply

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Economics (AREA)
  • Data Mining & Analysis (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Probability & Statistics with Applications (AREA)
  • Tourism & Hospitality (AREA)
  • Accounting & Taxation (AREA)
  • Finance (AREA)
  • Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Game Theory and Decision Science (AREA)
  • Evolutionary Computation (AREA)
  • Evolutionary Biology (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Water Supply & Treatment (AREA)
  • Quality & Reliability (AREA)
  • Operations Research (AREA)
  • Photovoltaic Devices (AREA)
  • General Health & Medical Sciences (AREA)
  • Public Health (AREA)
  • Primary Health Care (AREA)

Abstract

The invention belongs to the technical field of cost optimization of biomass power plants, and particularly relates to a device output prediction method based on a renewable energy micro-grid in an island operation mode of a biomass power plant, which comprises the steps of establishing a photovoltaic power prediction model for predicting short-term photovoltaic power generation in a probability density function form and obtaining a predicted probability density function; and (3) establishing a two-stage random optimization model, inputting the predicted probability density function into the two-stage random model to reduce renewable energy power and reduce load and minimum output cost of the cogeneration unit, and solving the two-stage random optimization problem to obtain the output condition of each device in a future period, so that decision making is facilitated in the past, and the capability of the power plant for resisting risks is enhanced.

Description

Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant
Technical field:
the invention belongs to the technical field of cost optimization of biomass power plants, and particularly relates to a device output prediction method of a renewable energy micro-grid based on an island operation mode of a biomass power plant.
The background technology is as follows:
carbon emissions are increasing due to urban and population growth around the world. Expert practitioners agree that these emissions are one of the main causes of climate change and global warming. The power industry is turning to renewable energy sources, such as photovoltaic power generation, and adopts different energy storage systems to realize the development of clean energy sources and make up for the scarcity of fossil fuels. However, the intermittent behavior of renewable resources creates some obstacles such as power fluctuations, adding additional backup units, and load shedding. Secondly, the rise of biomass power generation technology is becoming mature as a stable renewable energy power generation technology, and it is difficult to analyze the optimized operation of the island grid in a region centered on a certain biomass power plant. Micro grid technology is a promising solution, where different renewable energy sources and loads can be integrated into the grid. Under the condition that the capacity of the energy storage system is limited, a complex scheduling method is required for running the micro-grid in the island mode. However, previous researches on micro-grids are aimed at the operation problem, neglecting uncertainty of supply and demand or considering only the worst operation condition. However, uncertainty is an inherent attribute of the power system on the one hand; on the other hand, considering the worst operation situation results in unnecessarily increasing operation and planning costs, resulting in waste of resources.
The invention comprises the following steps:
in view of the above, the present invention provides a device output prediction method based on renewable energy microgrid in island operation mode of biomass power plant, comprising,
establishing a photovoltaic power prediction model for predicting short-term photovoltaic power generation capacity in a probability density function form, and obtaining a predicted probability density function;
and (3) establishing a two-stage random optimization model, inputting the predicted probability density function into the two-stage random optimization model, calculating the two-stage optimal cost, and enabling each device with the minimum objective function to output under the condition that the first and second two-stage constraint conditions are met, so as to realize the optimal scheduling decision of the current time of the renewable energy micro-grid to the future time device.
Further, the photovoltaic power prediction model is established by the following steps,
(1) Constructing binary probability distribution of observed data;
(1-1) acquiring historical photovoltaic output data;
(1-2) clustering historical photovoltaic output data into K clusters by using a K-means clustering algorithm to obtain photovoltaic power of each clustered cluster;
(1-3) establishing a two-dimensional histogram of each cluster (namely, probability under certain power at a certain moment) according to the photovoltaic power, verifying whether probability distribution in the two-dimensional histogram is optimal or not by adopting a Kolmogorov-Smirnov test (K-S test), and generating original probability density functions at different moments in the cluster K according to verification results;
(2) Construction of artificial neural networks
The method comprises the steps of (2-1) normalizing the weather parameters, the time constant n and the historical photovoltaic output data after clustering in the step 1), and inputting the normalized historical photovoltaic output data into an artificial neural network ANN to obtain predicted power data of each time of each cluster after de-normalization;
(2-2) generating a predicted probability density function from the predicted power data in the manner of (1-3) above.
Further, wherein, the binary probability distribution of the observed data is constructed, the specific steps are as follows,
step 1: n values of the observed data of the historical photovoltaic output data are input into a matrix S:
S＝[S 1 S 2 ......S N ] (1)
wherein S is N Is the photovoltaic power generation power P observed at the time constant N epsilon 1,2, … …, N WT Wherein the time constant indicates which hour.
Step 2: using a k-means clustering algorithm to represent data in a clustering form, and grouping data points with similar attributes into one cluster; in this step, the data in S is clustered into K clusters; since the photovoltaic power range is between 0 and its maximum recorded output value, the cluster number K should not be arbitrarily determined; in practice, careful selection should be made. The optimal cluster number K is a value satisfying a minimum error Sum of Squares (SSE), which is the sum of the squared distances between each element in the set and the corresponding average value of the set, as shown in formula (2):
wherein m is j Is S j Average value of S j Is a subgroup of S represented by cluster j.
Taking SSE in the formula (2) as a measure of the optimal cluster number k, calculating different k values, and determining the optimal k at a strong inflection point of DS; at the end of this step, the data in S will be replaced by the corresponding cluster, resulting in clustered data clusters S';
step 3: obtaining a two-dimensional (2D) histogram for each cluster after clustering; the two-dimensional histogram shows the frequencies that meet the photovoltaic power value P at the corresponding time constant h; the calculation method of the point (frequency) of each two-dimensional histogram of cluster j is as shown in formula (3):
wherein, xi is the frequency number of the power P when h is carried out on the jth cluster, and the value of N is 1-24 hours in the set of all the powers of the jth cluster;
after obtaining the two-dimensional histogram, for the observation at hThe Kolmogorov-Smirnov test (K-S test); the test is used to find the most appropriate probability distribution, effectively fitting the observed data.
Further, the K-S test in step 3 is to compare the sample data with the calculated theoretical distribution, and if the difference is not large, the data is considered to be in normal distribution, which isProbability density functions of (2); the difference D between the calculated theoretical distribution cumulative probability and the sample data cumulative probability n As shown in formula (4):
wherein F is 0 (x) Accumulating probabilities for theoretical distribution, F n (x) Accumulating probabilities for sample data;
if D n >D, not the most suitable probability distribution, needs to be recalculated, wherein D is an allowable difference value; by obtaining a two-dimensional histogram and an appropriate distribution function, a probability density function of the observed data is obtained every hour for each cluster.
Further, the optimization problem in the two-stage stochastic optimization model comprises two stages; the first phase is a variable of the current time h=0, and the second phase is a variable of the future possible cases of h=1, 2, …, n.
Further, the two-stage random optimization model establishment comprises the following steps,
s1, generating scenes by using the obtained probability density function, and calculating the occurrence probability of each scene;
s2, taking the probability value of the scene occurrence as a coefficient, establishing an objective function and constraint conditions of the two-stage optimization problem, and further generating a two-stage random optimization model.
Further, in step S1, each h has χ h The number of scenes is represented by equation (5) for the values of the possible photovoltaic powers:
further, in step S1, a probability value ρ of each scene occurrence τ Calculated by the following formula:
where τ represents the scene,is the photovoltaic power generation amount with a time constant of h in the situation tau.
Further, the objective function of the two-stage random optimization problem in step S2 is as follows,
wherein C is CHP,0 To optimize the output costs of the cogeneration unit, lambda, at the moment (first phase, h=0) cur And lambda (lambda) ENS The cost of renewable energy PV power reduction and load reduction, respectively.P cur,0 And P ENS,0 The reduced renewable energy photovoltaic power and the reduced load power at the optimization moment (at the first stage, h=0) respectively. H is the time constant e h=1, 2, …, n, where n is the future time constant considered in the optimization.And->And respectively generating cost, photovoltaic reduction power and load reduction power of the cogeneration unit in the future time constant h under the scene tau.
Further, the constraints of the two-stage stochastic optimization problem in step S2 include a first-stage constraint and a second-stage constraint, wherein,
the first-stage constraint conditions comprise power balance constraint, upper and lower limit constraint of electricity output of the cogeneration unit, electricity output constraint of the cogeneration unit section and charge and discharge constraint of an energy storage system;
the constraint conditions of the second stage comprise constraint of upper and lower limits of electric output of the cogeneration unit and constraint of the cogeneration unit l th The electric force constraint of the section and the charge and discharge constraint of the energy storage system.
According to the scheme, under the condition of uncertainty and load fluctuation of the renewable energy micro-grid in the island operation mode of the biomass power plant, uncertainty and fluctuation characteristics of the renewable energy micro-grid are described in a random scene mode, and according to the uncertainty and fluctuation characteristics of the renewable energy micro-grid, the unit output force required by not adding other energy supply equipment or minimum load reduction in island operation is found, and under the condition that the first and second two-stage constraint conditions are met, the output force of each equipment with the minimum objective function is enabled. And the optimal scheduling decision of the current time of the renewable energy micro-grid to the future time equipment is realized for achieving the purposes of energy conservation and emission reduction.
The specific embodiment is as follows:
the present invention will be further described in detail with reference to the drawings and examples, in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
The technical scheme of the invention provides a novel prediction model based on an Artificial Neural Network (ANN) aiming at uncertainty and fluctuation characteristics of a renewable energy micro-grid model, wherein the model represents a photovoltaic power prediction output value in the form of a probability density function. And a two-stage random optimization method is provided to minimize the running cost and reduce the load in the island mode, and in the provided random optimization method, the probability of the recent potential power supply, load and energy storage system capacity is considered, so that the optimal scheduling decision is made at the current moment.
The invention provides a cost prediction method of a renewable energy micro-grid based on an island operation mode of a biomass power plant, which comprises the steps of establishing a photovoltaic power prediction model for predicting short-term photovoltaic power generation in a probability density function mode, and obtaining a predicted probability density function; and establishing a two-stage random optimization model, and inputting the predicted probability density function into the two-stage random optimization model to calculate the two-stage optimal cost. The method comprises the following steps:
1 photovoltaic power prediction model
The proposed model aims at predicting the short term (.ltoreq.24 hours) photovoltaic power generation as a function of probability density. The probability density function that predicts future photovoltaic power values provides more data information about the future than constant value predictions. The model is obtained from two main parts: obtaining binary probability distribution of observed data and constructing an artificial neural network.
1.1 construction of binary probability distribution
In this section, a binary probability distribution of observed data will be obtained. It starts with importing observation data and ends with acquiring probability density functions.
Step 1: first, N values of the observation data of the photovoltaic power generation power are input into a matrix S shown in formula (1).
S＝[S 1 S 2 ......S N ] (1)
Wherein S is N Is the photovoltaic power generation power (P) observed at the time constant N epsilon 1,2, … …, N WT ) Wherein the time constant N indicates which hour.
Step 2: data is represented in clusters using a k-means clustering algorithm, grouping data points with similar attributes into one cluster. In this step, the data in S is clustered into a certain number of clusters. Since the photovoltaic power range is between 0 and its maximum recorded output value, the number of clusters should not be arbitrarily determined; the optimal cluster number is a value satisfying the least squares error sum (SSE). SSE is the sum of the squared distance between each element in the set and the corresponding average value for that set, as shown in equation (2).
SSE in the formula (2) is taken as a measure of the optimal cluster number k, different k values are calculated, and the optimal k is determined at the strong inflection point of DS.
Wherein m is j Is S j Average value of S j Is a subgroup of S represented by cluster j.
At the end of this step, the data in S will be replaced by the corresponding cluster, resulting in clustered data clusters S'.
Step 3: in this step, a two-dimensional (2D) histogram is obtained for each cluster after clustering. The two-dimensional histogram shows the frequencies that meet the photovoltaic power value P at the corresponding time constant h. The calculation method of the point (frequency) of each two-dimensional histogram of cluster j is as shown in formula (3):
wherein, xi is the frequency number of the power P when h is carried out on the jth cluster, and the value of N is 1-24 hours in the set of all the powers of the jth cluster;
after obtaining the two-dimensional histogram, for the observation at hThe Kolmogorov-Smirnov test (K-S test) was performed on the values of (a) the (b). The test is used to find the most appropriate probability distribution, effectively fitting the observed data.
The K-S test is to compare the sample data with the calculated theoretical distribution, and if the difference is not large, the data is considered to be subjected to normal distribution, namelyProbability density function of (a).
The difference between the calculated theoretical distribution cumulative probability and the sample data cumulative probability is shown as formula (4):
wherein F is 0 (x) Accumulating probabilities for theoretical distribution, F n (x) Accumulating probabilities for sample data;
if D n >D, which is not the most appropriate probability distribution, is recalculated, where D is the allowed difference.
By obtaining a two-dimensional histogram and an appropriate distribution function, a probability density function of the observed data is obtained every hour for each cluster. The probability density function for establishing different times is described above-as a standard, the predicted scene is one of the histograms described above.
1.2 construction of an Artificial neural network
Five input data are first prepared, the time step being n, namely the observed photovoltaic power (w), air pressure (KPa), topography (M), temperature (DEG C) and time constant (month M 0 And hours h 0 )。
Step 1: the photovoltaic output data will be clustered prior to entering the first step. The clustering of the data uses the same model as described in step 2 in the previous subsection (1.1). In the normalization step, the data is scaled to a range of-1 to speed up learning with tanh as the activation function.
And a second step of: normalized data will be input into the trained artificial neural network. The ANN consists of an input layer, several hidden layers and an output layer. The input layer consists of five input nodes, in addition to the time constant, there are photovoltaic output and weather parameters (temperature, air pressure and topography) for the last 24 hours. The sequence of each parameter is then fed into a set of long short-term memory (LSTM) layers to determine the time dependence of the parameter time sequence. The output of the output layer is between-1 and 1, representing the predicted data for n time steps.
And a third step of: in the de-normalization process, data is transferred from the (1 to-1) interval into the corresponding cluster j=1, 2,3, … …, k. The output of this step is the cluster value for the next n time steps.
Fourth step: given all the cluster values at h=1, 2,3, … …, n, a probability density function at each h constant number is generated using a cluster-probability density function map.
2 two stage random optimization
The goal of the optimization model is to take into account the uncertainty of the photovoltaic output, and to preferentially meet the critical load (critical load refers to the minimum load requirements that the microgrid must meet to ensure the reliability and safety of the microgrid). Mixed Integer Linear Programming (MILP) is utilized to solve the optimization problem. Decision variables and constraints thereof are defined first. Then, an objective function is set, taking into consideration the probability of occurrence of the second-stage scenario, with the objective of minimizing the cost and the cut load. The probability density function of the photovoltaic output obtained by the artificial neural network model is used as one of the inputs of the optimization model.
2.1 scene Generation phase
The optimization problem includes two phases; the first phase is a variable of the current moment (h=0), and the second phase is a variable of the future possible cases of h=1, 2, …, n. These future scenarios are the result of making a decision when h=0.
The second stage should take into account all possible cases to ensure that the next index (h>0) Is described. Firstly, generating a scene by using the probability density function obtained above:the number of scenes is set to n=k×m×24 scenes. The number of scenes can be set by the user, and a required random scene number can be selected, for example, 10 times (0, 1) intervals of random sampling are performed on the probability density function obtained by each clustered cluster in each hour, wherein the corresponding value of the (0, 1) interval is the probability of corresponding photovoltaic power, the corresponding abscissa is the corresponding power value, and the corresponding scene number is k×10×24. Each h has χ h The number of scenes is represented by equation (5) for the values of the possible photovoltaic powers:
for example, if there is only one value of possible photovoltaic power in each future h=1, 2, …, n, the number of future scenes is one.
The number of scenes is determined by equation (5), Ω being the set of scenes considered. Future scenes are predicted by the model based on artificial neural networks in section 1.2. Each scene τ, the probability of occurrence is ρ τ As coefficients of the target function (9) in the second stage. The probability is derived from the observed data. It is the conditional probability of an independent event, i.e. the value of the random number sampled in the above.
Wherein, the liquid crystal display device comprises a liquid crystal display device,is the photovoltaic power generation amount with a time constant of h in the situation tau.
2.2 objective function and constraint
Excess and insufficient power generation are two problems that island renewable energy micro-grids may face. Load shedding is one of measures to solve the problem of insufficient power generation, and the problem of excessive power generation is alleviated by reducing the power generated by renewable energy sources.
The two-stage stochastic optimization problem is to minimize the generation cost, photovoltaic power curtailment and unnecessary load shedding of the island renewable micro-grid. The biomass power plant usually adopts a cogeneration unit, and the running condition of the micro-grid is considered in the research. The power generation cost of the cogeneration unit of the biomass power plant is generally represented by a quadratic function, as shown in formula (7):
wherein P is CHP Is the output electric power of the cogeneration unit, and a, b and c are cost coefficients. However, since the MILP method is suitable for solving the linear problem, piecewise linear curve approximation is required for the power generation cost curve of the cogeneration unit to adapt to the MILP method, as shown in equation (8):
wherein C is cHP,t Is the power generation cost of the cogeneration unit at the time t, S l Is the slope of the linear portion l on the cost curve, Γ is a set of line segments, P CHP,l,t At time t at cost curve l th The output power of the segment.
The objective function of the two-stage random optimization problem is thus as follows:
wherein C is CHP,0 To optimize the output costs of the cogeneration unit, lambda, at the moment (first phase, h=0) cur And lambda (lambda) ENS The cost of renewable energy PV power reduction and load reduction, respectively. P (P) cur,0 And P ENS,0 The reduced renewable energy photovoltaic power and the reduced load power at the optimization moment (at the first stage, h=0) respectively. H is the time constant e h=1, 2, …, n, where n is the future time constant considered in the optimization.And->And respectively generating cost, photovoltaic reduction power and load reduction power of the cogeneration unit in the future time constant h under the scene tau. Each set of cascaded photovoltaic outputs that may occur from h=1 to h=n constitutes a scene τ. In each time constant h, there may be several possible photovoltaic output values, which depend on the number of samples of the scene generation phase.
2.3 first stage constraints
1) Power balance constraint
Wherein P is dis For discharging power of energy storage system, P ch And the charging power of the energy storage system. P (P) WT Is the output power of the photovoltaic unit. P (P) cur L is the power of the reduced photovoltaic for the surplus of generation period 0 P for total load demand ENS,0 To reduce the load power. The symbol 0 in this subsection indicates that these parameters belong to the first phase, i.e. the optimization moment h=0. The remaining constraints are as follows.
2) Upper and lower limit constraint of electric output of cogeneration unit
3) Combined heat and power unit l th Electrical force constraint of a segment
4) The charge and discharge constraint (formula (13) -formula (18)) of the energy storage system is shown
U ch,0 +U dis,0 ≤1 (13)
Wherein U is ch,0 And U dis,0 Respectively, representing the charge and discharge electrical states of the energy storage system, indicated by 0, 1.
Wherein SOC is h-1 For the residual charge quantity of the previous moment eta ch And eta dis Respectively charge and discharge efficiency, P ch,h And P dis,h For the charge-discharge power at time h, E cap For energy storage system capacity, Δh is every time step (1 hour).
SOC min ≤SOC 0 ≤SOC max (17)
SOC 0-1 When the power generation of the power generation element (50) (18) is excessive, the power generation output constraint of the renewable energy source can be reduced
0≤P cur,0 ≤P WT,0 (19) The period of insufficient power generation reduces the non-critical load power constraint (formula (20) -formula (21)
0≤P ENS,0 ≤L Non-C,0 (20)
L 0 ＝L C +L Non-C,0 (21)
Wherein L is C Is critical load, L Non-C ， 0 Is a non-critical load, L 0 As the total load
2.4 constraints of the second stage
In the second stage, all future scenes should be considered simultaneouslyTo ensure that the solution is always viable. The constraint of power balance is expressed by a formula.
Wherein, the liquid crystal display device comprises a liquid crystal display device,and generating power of the cogeneration unit at the moment h for the line segment l in the scene tau. />And->The discharging power and the charging power of the energy storage system are respectively. />And L h Is the reduced photovoltaic power, photovoltaic output and load demand at time h.
The remaining constraints in the second stage are as follows:
1) Upper and lower limit constraint of electric output of cogeneration unit
2) Combined heat and power unit l th Electrical force constraint of a segment
3) The charge and discharge constraint (formula (25) -formula (27)) of the energy storage system is shown
The constraint of the renewable energy source power generation output can be reduced when the power generation is excessive
The period of insufficient power generation reduces the non-critical load power constraint (formula (32) -formula (33))
In the method, in the process of the invention,for the electric power of the cogeneration unit at time h in scene tau, < >>Maximum rated electric output power of cogeneration unit,/->And->Representing the discharge and charge states, respectively, of the energy storage system, represented by 0, 1.And->The SOCs at times h, h-1 and n in scene τ, respectively. The output of the schedulable generator is limited by constraints (11), (12), (23) and (24). The energy storage system discharge and charge states are mutually exclusive, represented by constraints (13) and (25). The maximum charge-discharge power of the energy storage system is limited by constraints (14), (15), (26), (27). At some point, just when the SOC reaches a maximum limit, excess power may be generated. Therefore, constraints (19) and (31) will limit the reduction of the photovoltaic output. SOC is defined by (16) and (28) and is limited by the maximum and minimum residual charge amounts in (17) and (29). The initial value and the end value of SOC are determined by (18) and (30). Non-critical load shedding is limited to (20) and (32) to support balanced power. The total load is determined by the accumulation of critical and non-critical loads in equations (21) and (33).
So far, the two-stage random optimization model of the renewable energy micro-grid under the island operation mode of the biomass power plant is built, and the two-stage optimal cost of the renewable energy micro-grid under the island operation mode of the specific area can be solved according to the data required by the model, so that the power output condition of each device in the future period is obtained, the decision making in the past is facilitated, the current production and life are guided, and the risk resistance capability of the power plant is enhanced. The prediction method in the scheme is also applicable to wind power energy sources.
The foregoing description of the preferred embodiments of the invention is not intended to be limiting, but rather is intended to cover all modifications, equivalents, and alternatives falling within the spirit and principles of the invention.

Claims (10)

1. The equipment output prediction method based on the renewable energy micro-grid under the island operation mode of the biomass power plant is characterized by comprising the following steps of: comprising the steps of (a) a step of,
establishing a photovoltaic power prediction model for predicting short-term photovoltaic power generation capacity in a probability density function form, and obtaining a predicted probability density function;
and (3) establishing a two-stage random optimization model, inputting the predicted probability density function into the two-stage random optimization model, and solving the two-stage random optimization model by taking the minimum cost as a target to obtain the output of each device in the future period.
2. The method for predicting the equipment output of the renewable energy microgrid based on the island operation mode of the biomass power plant according to claim 1, wherein the method comprises the following steps of: the photovoltaic power prediction model is established through the following steps that 1, binary probability distribution of observation data is established;
1.1, acquiring historical photovoltaic output data;
1.2, clustering historical photovoltaic output data into K clusters by using a K-means clustering algorithm to obtain photovoltaic power of each clustered cluster;
1.3, building a two-dimensional histogram of each cluster according to the photovoltaic power; verifying whether the probability distribution in the two-dimensional histogram is optimal or not by adopting a Kolmogorov-Smirnov test, and generating original probability density functions at different moments in the cluster K according to a verification result;
2 construction of an Artificial neural network
2.1, normalizing the weather parameters, the future time n and the historical photovoltaic output data after clustering in the step 1), inputting the normalized historical photovoltaic output data into an artificial neural network, and removing the predicted power data of each time of each cluster obtained after normalization;
2.2 generating a predicted probability density function from the predicted power data in the manner of step 1.3 above.
3. The method for predicting the equipment output of the renewable energy microgrid based on the island operation mode of the biomass power plant according to claim 2, wherein the method is characterized by comprising the following steps of constructing binary probability distribution of observed data,
step 1: n values of the observed data of the historical photovoltaic output data are input into a matrix S:
S＝[S 1 S 2 ......S N ] (1)
wherein S is N Is the photovoltaic power generation power P observed at the time N epsilon 1,2, … …, N WT Wherein time indicates what number of hours;
step 2: clustering the data in S into K clusters by using a K-means clustering algorithm; where the cluster number K is a value that satisfies a minimum error squared sum SSE, which is the sum of the squared distance between each element in the set and the corresponding average value for the set, as shown in equation (2):
wherein m is j Is S j Average value of S j Is a subgroup of S represented by cluster j;
taking SSE in the formula (2) as a measure of the optimal cluster number k, calculating different k values, and determining the optimal k at a strong inflection point of DS; at the end of this step, the data in S will be replaced by the corresponding cluster, resulting in clustered data clusters S';
step 3: obtaining a two-dimensional histogram for each cluster after clustering; the two-dimensional histogram shows the frequencies that meet the photovoltaic power value P at the corresponding time h; the calculation method of the frequency represented by the points of each two-dimensional histogram of cluster j is as shown in formula (3):
wherein, xi is the frequency number of the power P when h is carried out on the jth cluster, and the value of N is 1-24 hours in the set of all the powers of the jth cluster;
after obtaining the two-dimensional histogram, for the observation at hThe values in (2) are subjected to Kolmogorov-Smirnov test, the most suitable probability distribution is found, and the observed data is fitted.
4. The method for predicting equipment output of renewable energy micro-grid based on island operation mode of biomass power plant as claimed in claim 3, wherein the K-S test in step 3 is to compare sample data with calculated theoretical distribution, and if the difference is not large, the data can be considered to be subjected to normal distribution, which isIs provided, in particular,
the difference D between the calculated theoretical distribution cumulative probability and the sample data cumulative probability n As shown in formula (4):
wherein F is 0 (x) Accumulating probabilities for theoretical distribution, F n (x) Accumulating probabilities for sample data;
if D n >D, not the most suitable probability distribution, needs to be recalculated, wherein D is an allowable difference value; by obtaining a two-dimensional histogram and an appropriate distribution function, a probability density function of the observed data is obtained every hour of each cluster.
5. The method for predicting the equipment output of the renewable energy microgrid based on the island operation mode of the biomass power plant according to claim 1, wherein the method comprises the following steps of: the optimization problem in the two-stage random optimization model comprises two stages, wherein the first stage is a variable with h=0 at the current moment; the second phase is a variable of h=1, 2, …, n, which is a possible future time instance.
6. The method for predicting the equipment output of the renewable energy microgrid based on the island operation mode of the biomass power plant according to claim 5, wherein the two-stage random optimization model is established and comprises the following steps of,
s1, generating scenes by using the obtained probability density function, and calculating the occurrence probability value of each scene;
s2, taking the probability value of each scene as a coefficient, establishing an objective function of the two-stage optimization problem, setting constraint conditions, and further generating a two-stage random optimization model.
7. The method for predicting equipment output of renewable energy microgrid based on island operation mode of biomass power plant according to claim 6, wherein in step S1, each h has χ h The number of scenes is represented by equation (5) for the values of the possible photovoltaic powers:
8. the method for predicting equipment output of renewable energy microgrid based on island operation mode of biomass power plant according to claim 6, wherein in step S1, probability value ρ of each scene occurrence is τ Calculated by the following formula:
where τ represents the scene,is the photovoltaic power generation amount with a time constant of h in the situation tau.
9. The method for predicting the equipment output of the renewable energy microgrid based on the island operation mode of the biomass power plant according to claim 6, wherein the objective function of the two-stage random optimization problem in the step S2 is as follows,
wherein C is CHP,0 To optimize the output costs of the cogeneration unit, lambda, at the moment (first phase, h=0) cur And lambda (lambda) ENS Cost of reducing renewable energy source PV power and load respectively, P Cur,0 And P ENS,0 The reduced renewable energy photovoltaic power and the reduced load power at the moment of optimization (first stage, h=0), respectively, H being the time constant e h=1, 2, …, n, where n is the future time constant considered in the optimization,and->And respectively generating cost, photovoltaic reduction power and load reduction power of the cogeneration unit in the future time constant h under the scene tau.
10. The method for predicting equipment output of the renewable energy micro-grid based on the island operation mode of the biomass power plant according to claim 6, wherein the constraint condition of the two-stage random optimization problem in the step S2 comprises a first-stage constraint condition and a second-stage constraint condition, wherein,
the first-stage constraint conditions comprise power balance constraint, upper and lower limit constraint of electricity output of the cogeneration unit, electricity output constraint of the cogeneration unit section and charge and discharge constraint of an energy storage system;
the constraint conditions of the second stage comprise constraint of upper and lower limits of electric output of the cogeneration unit and constraint of the cogeneration unit l th The electric force constraint of the section and the charge and discharge constraint of the energy storage system.
CN202310784504.6A 2023-06-29 2023-06-29 Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant Pending CN116843066A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310784504.6A CN116843066A (en) 2023-06-29 2023-06-29 Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310784504.6A CN116843066A (en) 2023-06-29 2023-06-29 Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant

Publications (1)

Publication Number Publication Date
CN116843066A true CN116843066A (en) 2023-10-03

Family

ID=88162832

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310784504.6A Pending CN116843066A (en) 2023-06-29 2023-06-29 Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant

Country Status (1)

Country Link
CN (1) CN116843066A (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117312899A (en) * 2023-11-30 2023-12-29 国网浙江省电力有限公司 Photovoltaic output typical scene generation method, system and storage medium
CN117591814A (en) * 2024-01-19 2024-02-23 北京志翔科技股份有限公司 Data restoration method, device and equipment based on photovoltaic envelope

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117312899A (en) * 2023-11-30 2023-12-29 国网浙江省电力有限公司 Photovoltaic output typical scene generation method, system and storage medium
CN117591814A (en) * 2024-01-19 2024-02-23 北京志翔科技股份有限公司 Data restoration method, device and equipment based on photovoltaic envelope
CN117591814B (en) * 2024-01-19 2024-06-07 北京志翔科技股份有限公司 Data restoration method, device and equipment based on photovoltaic envelope

Similar Documents

Publication Publication Date Title
Mirzapour et al. A new prediction model of battery and wind-solar output in hybrid power system
Maulik et al. Optimal operation of droop-controlled islanded microgrids
CN109325608B (en) Distributed power supply optimal configuration method considering energy storage and considering photovoltaic randomness
Logenthiran et al. Short term generation scheduling of a microgrid
Zargar et al. Development of a Markov-chain-based solar generation model for smart microgrid energy management system
CN116843066A (en) Equipment output prediction method based on renewable energy micro-grid in island operation mode of biomass power plant
Wang et al. Optimal planning of stand-alone microgrids incorporating reliability
CN109034587B (en) Active power distribution system optimal scheduling method for coordinating multiple controllable units
Abunima et al. Two-stage stochastic optimization for operating a renewable-based microgrid
CN110245794B (en) Flexibility-considered double-layer optimization method for central fire storage capacity in multi-energy convergence
Zhang et al. Design and optimal sizing of hybrid PV/wind/diesel system with battery storage by using DIRECT search algorithm
Hou et al. Energy storage system optimization based on a multi-time scale decomposition-coordination algorithm for wind-ESS systems
Shu et al. Optimal sizing of energy storage system for wind power plants
CN117060474A (en) Scheduling method, system, equipment and storage medium of new energy charging station
Esmaeili et al. Market-oriented optimal control strategy for an integrated energy storage system and wind farm
CN114336762A (en) Day-ahead scheduling energy storage configuration optimization method for wind-solar power generation and power grid load fluctuation
CN113708418A (en) Micro-grid optimization scheduling method
CN113435659A (en) Scene analysis-based two-stage optimized operation method and system for comprehensive energy system
Hong et al. Short-term real-power scheduling considering fuzzy factors in an autonomous system using genetic algorithms
Mohiti et al. Frequency‐constrained energy and reserve scheduling in wind incorporated low‐inertia power systems considering vanadium flow redox batteries
Muttaqi et al. An effective power dispatch strategy to improve generation schedulability by mitigating wind power uncertainty with a data driven flexible dispatch margin for a wind farm using a multi-unit battery energy storage system
Paliwal et al. Short-term optimal energy management in stand-alone microgrid with battery energy storage
CN114204552A (en) Comprehensive energy scheduling method for intelligent park
CN114498690A (en) Multi-element composite energy storage optimal configuration method supporting large-scale renewable energy consumption
Gnanaprakasam et al. An efficient MFM-TFWO approach for unit commitment with uncertainty of DGs in electric vehicle parking lots

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