CN111680815A - BP neural network-based micro-grid hierarchical optimization reconstruction method - Google Patents

BP neural network-based micro-grid hierarchical optimization reconstruction method Download PDF

Info

Publication number
CN111680815A
CN111680815A CN202010288722.7A CN202010288722A CN111680815A CN 111680815 A CN111680815 A CN 111680815A CN 202010288722 A CN202010288722 A CN 202010288722A CN 111680815 A CN111680815 A CN 111680815A
Authority
CN
China
Prior art keywords
power
node
reconstruction
micro
neural network
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
CN202010288722.7A
Other languages
Chinese (zh)
Inventor
李矗
林加阳
易永利
王亮
周伟豪
向魁
李武
吴堃铭
周震宇
陈民铀
陈达
颜军敏
乐雪清
叶正策
万晓
雷欢
钱碧甫
滕志豪
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.)
Chongqing University
State Grid Corp of China SGCC
State Grid Zhejiang Electric Power Co Ltd
Wenzhou Power Supply Co of State Grid Zhejiang Electric Power Co Ltd
Original Assignee
Chongqing University
State Grid Corp of China SGCC
State Grid Zhejiang Electric Power Co Ltd
Wenzhou Power Supply Co of State Grid Zhejiang Electric Power Co Ltd
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 Chongqing University, State Grid Corp of China SGCC, State Grid Zhejiang Electric Power Co Ltd, Wenzhou Power Supply Co of State Grid Zhejiang Electric Power Co Ltd filed Critical Chongqing University
Priority to CN202010288722.7A priority Critical patent/CN111680815A/en
Publication of CN111680815A publication Critical patent/CN111680815A/en
Pending legal-status Critical Current

Links

Images

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
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • 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
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/04—Constraint-based CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2113/00—Details relating to the application field
    • G06F2113/04—Power grid distribution networks

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Business, Economics & Management (AREA)
  • Theoretical Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • General Physics & Mathematics (AREA)
  • Strategic Management (AREA)
  • Evolutionary Computation (AREA)
  • Economics (AREA)
  • Software Systems (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Development Economics (AREA)
  • Artificial Intelligence (AREA)
  • General Engineering & Computer Science (AREA)
  • Biophysics (AREA)
  • Operations Research (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Educational Administration (AREA)
  • General Business, Economics & Management (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • Computing Systems (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Game Theory and Decision Science (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Data Mining & Analysis (AREA)
  • Marketing (AREA)
  • Mathematical Physics (AREA)
  • Geometry (AREA)
  • Computer Hardware Design (AREA)
  • Medical Informatics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Supply And Distribution Of Alternating Current (AREA)

Abstract

The invention discloses a BP neural network-based microgrid hierarchical optimization reconstruction method, which comprises the following steps of S1: establishing a micro-grid optimization reconstruction model, and establishing a target function and constraint conditions of micro-grid optimization reconstruction; s2: establishing a BP neural network model, and training and testing the neural network; s3: by utilizing a hierarchical optimization thought, in the reconstruction of the micro-grid, the method is divided into a first secondary optimization process according to whether load flow calculation is carried out or not, and a trained BP neural network is used for replacing the load flow calculation in the second-stage optimization process; s4: introducing a comprehensive evaluation method, constructing a comprehensive evaluation function, carrying out comprehensive evaluation on the output result of the BP network, and selecting an optimal reconstruction scheme. The invention has the advantages that: in the reconstruction process, the BP neural network is trained offline, the on-off state is input online, the network loss is directly output, the node power and voltage results are balanced, the calculation time is saved, and the reconstruction efficiency of the microgrid is obviously improved.

Description

BP neural network-based micro-grid hierarchical optimization reconstruction method
Technical Field
The invention relates to the technical field of microgrid optimization reconstruction, in particular to a microgrid hierarchical optimization reconstruction method based on a BP (Back propagation) neural network.
Background
In recent years, with the development of new energy power generation, attention has been paid to micro grids. The micro-grid is a small power generation and distribution system organically integrating a distributed power supply, a load, an energy storage device, a current transformer and a monitoring protection device, and has two modes of island operation and grid-connected operation. The micro-grid is likely to break down in the operation process, the switching state in the micro-grid needs to be adjusted at the moment, and after the micro-grid is reconstructed, the load can be balanced, the grid loss can be effectively reduced, the electric energy quality can be improved, and the like, so that the reliability and the safety of the micro-grid are improved. With the development of network reconstruction technology, researchers begin to research the accuracy and speed of network reconstruction while paying attention to a network reconstruction algorithm.
In the existing micro-grid reconstruction algorithm, the number of variables is increased along with the increase of the number of network nodes, so that the calculation amount and the calculation time are increased. In addition, some intelligent algorithms can only find local optimum and cannot find global optimum solutions, so that whether the final reconstruction result reaches an optimum state cannot be guaranteed. In the reconstruction process of the micro-grid, a hierarchical optimization thought is introduced, and a BP neural network is used for replacing tidal current calculation, so that the reconstruction precision is guaranteed, the tidal current calculation pressure during the reconstruction of the micro-grid is reduced, and the reconstruction efficiency is greatly improved.
Disclosure of Invention
In view of the above defects in the prior art, the present invention aims to provide a micro-grid hierarchical optimization reconstruction method based on a BP neural network, in the micro-grid reconstruction process, matpower is used to calculate and obtain BP neural network training data, the neural network is trained offline, and a hierarchical optimization idea is introduced to calculate network loss, balance node power offset, voltage deviation, etc., so as to save calculation time and significantly improve micro-grid reconstruction efficiency.
In order to solve the technical problems, the invention is realized by the following technical scheme: a micro-grid hierarchical optimization reconstruction method based on a BP neural network comprises the following steps:
s1: establishing a micro-grid optimization reconstruction model, and establishing a target function and constraint conditions of micro-grid optimization reconstruction;
s2: establishing a BP neural network model, and training and testing the neural network;
s3: by utilizing a hierarchical optimization thought, in the reconstruction of the micro-grid, the optimization process which does not involve load flow calculation is put into the first-stage processing, and the optimization process which involves load flow calculation is taken as the second-stage optimization processing; only the scheme meeting the first-stage optimization condition is subjected to second-stage optimization processing, and a trained BP neural network is used for replacing load flow calculation in the second-stage optimization process;
s4: introducing a comprehensive evaluation method, constructing a comprehensive evaluation function, carrying out comprehensive evaluation on the output result of the BP network, and selecting an optimal reconstruction scheme.
Preferably, in step S1, under the constraint condition, the objective function of the microgrid optimization reconstruction is as follows:
s11: the micro-grid reconstruction needs to meet the goal of minimum load shedding amount, and the objective function is as follows:
Figure BDA0002449579200000021
wherein i ∈ omega is a node set for removing load after reconstruction, and SiRepresenting the load amount corresponding to the node i;
the stable operation of the reconstructed microgrid needs to meet the following constraint conditions:
1) balancing node power constraints:
Ptmin≤Pt≤Ptmax
in the formula: ptActive power adjustable for the balancing node t; ptmaxThe upper limit of the active power can be adjusted for the node t; ptminThe lower limit of the active power can be adjusted for the node t;
2) branch power constraint:
PBj≤PBjmax
in the formula: pGxThe power generation power of a micro power source X in the micro power grid is represented, wherein the X represents the number of the micro power sources reserved after reconstruction; pLiLoad active power reserved for the nodes i after the microgrid is reconstructed, wherein N represents the number of the nodes reserved after the reconstruction;
3) and power balance constraint:
Figure BDA0002449579200000031
in the formula: pGxThe power generation power of a micro power source X in the micro power grid is represented, wherein the X represents the number of the micro power sources reserved after reconstruction; pLiLoad active power reserved for the nodes i after the microgrid is reconstructed, wherein N represents the number of the nodes reserved after the reconstruction;
4) micro-power source power generation constraint:
PGmin≤PG≤PGmax
in the formula: pGThe total generated power of the micro power supply after the micro power grid is reconstructed; pGminThe lower limit of the generated power in the micro-grid; pGmaxThe upper limit of the generated power in the micro-grid;
5) node voltage constraint:
Uimin≤Ui≤Uimax
in the formula: u shapeiIs the voltage magnitude of node i; u shapeiminIs the lower voltage limit of node i; u shapeimaxIs the upper voltage limit of node i.
Preferably, the BP neural network in step S2 is constructed as follows:
s21: in order to improve the training precision, the neural network model constructed by the invention contains 2 hidden layers, each hidden layer
The node number of each hidden layer is determined according to the following formula:
Figure BDA0002449579200000041
in the formula, h is the number of nodes of the hidden layer, z and v are the number of nodes of the input layer and the output layer respectively, and c is an adjusting constant between 1 and 10.
Preferably, the training test process of the BP neural network in step S2 is as follows:
s22: calculating the power flows of the micro-grid corresponding to different switch states by using a Matpower toolkit, wherein the power flows comprise balance node power PtNetwork loss L, node maximum voltage UmaxAnd a minimum voltage Umin(ii) a Collecting the data;
s23: in the collected data, 80% of the data is selected as a training set at will, and a neural network toolbox in Matlab is used for training;
s24: and aiming at the BP neural network obtained by training, introducing the rest 20% of the BP neural network as test set data into the BP neural network, comparing the output result with the data in the test set, and judging that the BP neural network meets the requirements.
Preferably, the step S3 applies the hierarchical optimization idea to the microgrid reconstruction as follows:
s31: in the first-stage optimization process, processing an objective function and constraint which do not involve any load flow calculation; considering the balance between the generated energy and the load quantity, performing integer programming to obtain a load switch state combination meeting the balance condition, and sequencing according to the ascending order of the power supply load shedding quantity to obtain a switch combination solution set D;
s32: in the second-stage optimization process, processing an objective function and constraint related to load flow calculation, substituting the switch combination solution set D obtained by the first-stage optimization into a BP neural network one by one for prediction, and obtaining a prediction result;
s33: judging whether the prediction result meets the relevant constraint of the stable operation of the micro-grid or not for the load shedding amount; if only one group of switch combination is satisfied, the switch combination is the optimal solution of the micro-grid reconstruction; on the contrary, if the output results of a plurality of groups of switch combinations meet the constraint, selecting the optimal solution of the micro-grid reconstruction according to the comprehensive evaluation method; and if all the combinations are not satisfied, selecting the next load shedding amount to repeat the steps until the optimal solution is found.
Preferably, the comprehensive evaluation method in step S4 is as follows:
s41: if the output results of a plurality of groups of switch combinations meet the constraint condition in the same load cutting amount, a plurality of groups of corresponding network loss, voltage deviation and power deviation outputs can be obtained; setting reference values of network loss, voltage deviation and power deviation, normalizing the output results of each group of switches, and selecting corresponding weight coefficients k according to the preference of a decision makeriAnd selecting the switch combination with the minimum comprehensive evaluation function value as the optimal solution by utilizing the comprehensive evaluation function.
Preferably, the comprehensive evaluation function in step S4 is constructed as follows:
s42: the reconstruction of the micro-grid needs to meet the requirements when the load shedding amount is the same and the results all meet the constraint conditions
The objective function of the minimum comprehensive evaluation function value is as follows:
Figure BDA0002449579200000051
in the formula: k is a radical of1，k2，k3Weight coefficients for network loss, voltage deviation and power deviation of balance node, and 0 < k1,k2,k3＜1，k1+k2+k3＝1；
Figure BDA0002449579200000052
Respectively normalizing values of network loss, voltage deviation and power deviation of a balance node;
s43: before comprehensive evaluation, the output result is firstly normalized and calculated
The method comprises the following steps:
Figure BDA0002449579200000053
in the formula: l is the system network loss value, L*Is a set network loss reference value; Δ U is the voltage deviation, which is the difference between the highest voltage and the lowest voltage of the node in the whole system, U*Is a system reference voltage; delta PtFor power deviation, it means the deviation of the power of the balance node from the set reference power, Pt *The reference power of the balance node is the average value of the upper and lower limits of the power of the balance node; wherein,
Figure BDA0002449579200000061
in the formula: pBj、QBj、RBjRespectively, active power, reactive power and resistance corresponding to branch j, wherein QBjThe method is obtained according to the load power factor and the active power; u shapeBjThe voltage at the j end of the branch thereof; r isjFor the branch state after corresponding reconstruction, 1 represents connection, and 0 represents disconnection; j is the number of branches contained in the system.
Preferably, in step S3, the exception handling process is as follows:
s34: if the output results of all switch combinations in the switch state set D do not meet the constraint condition, indicating that the fault reconstruction is abnormal, and entering an abnormal processing mechanism; in all the switch combinations which do not satisfy the constraint condition, the node voltage deviation and the balanced node power threshold value of each switch combination are comprehensively calculated to be more limited Y, and the method is as follows:
Figure BDA0002449579200000062
in the formula, w1，w2Is a weight coefficient of 0<w1，w2<1，w1+w 21 is ═ 1; i is the minimum voltage node, dUiminIs the lower out-of-limit absolute value of the minimum voltage, UiminIs the lower limit value of the node voltage; j is the maximum voltage node, dUjmaxIs the upper out-of-limit absolute value of the maximum voltage, UjmaxIs a node voltage upper limit value; t is a balance node, dPtminTo balance the absolute value of the off-limit power at the node, PtminTo balance node power lower limit, dPtmaxTo balance the absolute value of the out-of-limit power on the node, PtmaxBalancing the node power upper limit value; and selecting the switch combination with the minimum comprehensive limit Y as the optimal reconstruction output solution.
Compared with the prior art, the invention has the advantages that:
(1) by introducing the hierarchical optimization idea, the load flow calculation pressure during the reconstruction of the micro-grid can be reduced.
(2) By utilizing the BP neural network, offline training is realized, the on-off state is input online, the result is directly output, network reconstruction load flow calculation is not performed, and the calculation time is saved.
Drawings
Fig. 1 is a schematic flow diagram of a micro-grid hierarchical optimization reconstruction method based on a BP neural network.
Fig. 2 is a structural schematic diagram of an improved 33-node microgrid system in a microgrid hierarchical optimization reconstruction method based on a BP neural network.
Fig. 3 is a BP neural network model in a microgrid hierarchical optimization reconstruction method based on a BP neural network.
Fig. 4 is a training mean square error diagram of a BP neural network model in a microgrid hierarchical optimization reconstruction method based on a BP neural network.
FIG. 5 is a training data fitting graph of a BP neural network model in the microgrid hierarchical optimization reconstruction method based on the BP neural network.
FIG. 6 is a BP neural network test output error diagram in a micro-grid hierarchical optimization reconstruction method based on a BP neural network.
Detailed Description
Reference will now be made in detail to embodiments of the present invention, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the drawings are illustrative and intended to be illustrative of the invention and are not to be construed as limiting the invention.
As shown in fig. 1-6; a micro-grid hierarchical optimization reconstruction method based on a BP neural network comprises the following steps:
s1: establishing a micro-grid optimization reconstruction model, and establishing a target function and constraint conditions of micro-grid optimization reconstruction;
s2: establishing a BP neural network model, and training and testing the neural network;
s3: by utilizing the hierarchical optimization thought, in the reconstruction of the micro-grid, the optimization process which does not involve load flow calculation is put into the first-stage processing, and the optimization process which involves load flow calculation is taken as the second-stage optimization processing. Only the scheme meeting the first-stage optimization condition is subjected to second-stage optimization processing, and a trained BP neural network is used for replacing load flow calculation in the second-stage optimization process;
s4: introducing a comprehensive evaluation method, constructing a comprehensive evaluation function, carrying out comprehensive evaluation on the output result of the BP network, and selecting an optimal reconstruction scheme.
More specifically, in the microgrid optimization reconstruction, the step 1) comprises the following steps:
s11: the micro-grid reconstruction needs to meet the goal of minimum load shedding amount, and the objective function is as follows:
Figure BDA0002449579200000081
wherein i ∈ omega is a node set for removing load after reconstruction, and SiRepresenting the load corresponding to node i.
The stable operation of the reconstructed microgrid needs to meet the following constraint conditions:
1) balancing node power constraints:
Ptmin≤Pt≤Ptmax；(2)
in the formula: ptActive power adjustable for the balancing node t; ptmaxThe upper limit of the active power can be adjusted for the node t; ptminThe lower limit of the active power can be adjusted for the node t.
2) Branch power constraint:
PBj≤PBjmax；(3)
in the formula: pBjIs the active power flowing through branch j; pBjmaxThe upper limit of the active power transmission for branch j.
3) And power balance constraint:
Figure BDA0002449579200000091
in the formula: pGxThe power generation power of a micro power source X in the micro power grid is represented, wherein the X represents the number of the micro power sources reserved after reconstruction; pLiAnd load active power reserved for the node i after the microgrid is reconstructed, wherein N represents the number of the nodes reserved after the reconstruction.
4) Micro-power source power generation constraint:
PGmin≤PG≤PGmax；(5)
in the formula: pGThe total generated power of the micro power supply after the micro power grid is reconstructed; pGminThe lower limit of the generated power in the micro-grid; pGmaxThe upper limit of the generated power in the microgrid.
5) Node voltage constraint:
Uimin≤Ui≤Uimax；(6)
in the formula: u shapeiIs the voltage magnitude of node i; u shapeiminIs the lower voltage limit of node i; u shapeimaxIs the upper voltage limit of node i.
Step 2) in the construction of BP neural network, training and testing steps are as follows,
s21: in order to improve the training precision, the neural network model constructed by the invention comprises 2 hidden layers, and the node number of each hidden layer is determined according to the following formula:
Figure BDA0002449579200000101
in the formula, h is the number of nodes of the hidden layer, z and v are the number of nodes of the input layer and the output layer respectively, and c is an adjusting constant between 1 and 10.
S22: calculating the power flows of the micro-grid corresponding to different switch states by using a Matpower toolkit, wherein the power flows comprise balance node power PtNetwork loss L, node maximum voltage UmaxAnd a minimum voltage UminAnd so on. The data is collected.
S23: of the collected data, 80% of the data were arbitrarily selected as a training set, and the remaining 20% were selected as a test set. For data in a training set, the on-off state of a three-level load is used as BP neural network input quantity and is led into an input layer; and (4) introducing the balance node power, the network loss, the maximum voltage and the minimum voltage of the node and the like calculated by the Matpower toolkit into an output layer as output quantities. And selecting the number of hidden layers and the number of nodes corresponding to each hidden layer by using a neural network toolbox in Matlab, and then setting the minimum error, the maximum training times, the learning rate and the transfer function of a training target to train within the maximum training range.
S24: and aiming at the trained BP neural network, introducing the switch state in the test set data into an input layer, operating the BP neural network to obtain output results of balanced node power, network loss, node maximum voltage, node minimum voltage and the like, comparing the output results with the data in the test set, calculating the error of each group of results, and if the error of each group of results is within a given error range, the trained BP neural network meets the requirement.
Step 3), after the BP neural network is obtained, applying the hierarchical optimization idea to the microgrid reconstruction process as follows:
s31: in the first-stage optimization process, the objective function and the constraint which do not involve any load flow calculation are processed. And considering balance between the generated energy and the load quantity, performing integer programming to obtain a load switch state combination meeting balance conditions, and sequencing according to the ascending order of the power supply load shedding quantity to obtain a switch combination solution set D.
S32: in the second-stage optimization process, objective functions and constraints related to load flow calculation are processed. And substituting the switch combination solution set D obtained by the first-stage optimization into a BP neural network one by one for prediction to obtain a prediction result.
S33: and judging whether the prediction result meets the relevant constraint of the stable operation of the micro-grid or not for the load shedding amount. And if only one group of switch combination is satisfied, the switch combination is the optimal solution for the reconstruction of the microgrid. On the contrary, if the output results of a plurality of groups of switch combinations meet the constraint, the optimal solution of the micro-grid reconstruction is selected according to the comprehensive evaluation method. And if all the combinations are not satisfied, selecting the next load shedding amount to repeat the steps until the optimal solution is found.
S34: and if the output results of all the switch combinations in the switch state set D do not meet the constraint condition, indicating that the fault reconstruction is abnormal, and entering an abnormal processing mechanism. In all the switch combinations which do not satisfy the constraint condition, the node voltage deviation and the balanced node power threshold value of each switch combination are comprehensively calculated to be more limited Y, and the method is as follows:
Figure BDA0002449579200000111
in the formula, w1，w2Is a weight coefficient of 0<w1，w2<1，w1+w 21. i is the minimum voltage node, dUiminIs the lower out-of-limit absolute value of the minimum voltage, UiminIs the lower limit value of the node voltage; j is the maximum voltage node, dUjmaxIs the upper out-of-limit absolute value of the maximum voltage, UjmaxIs a node voltage upper limit value; t is a balance node, dPtminTo balance the absolute value of the off-limit power at the node, PtminTo balance node power lower limit, dPtmaxTo balance the absolute value of the out-of-limit power on the node, PtmaxTo balance the node power upper limit. And selecting the switch combination with the minimum comprehensive limit Y as the optimal reconstruction output solution.
Step 4) the comprehensive evaluation process is as follows:
s41: if the output results of a plurality of groups of switch combinations meet the constraint condition in the same load cutting amount, a plurality of groups of corresponding network loss, voltage deviation and power deviation outputs can be obtained. Setting reference values of network loss, voltage deviation and power deviation, normalizing the output results of each group of switches, and selecting corresponding weight coefficients k according to the preference of a decision makeriAnd selecting the switch combination with the minimum comprehensive evaluation function value as the optimal solution by utilizing the comprehensive evaluation function.
S42: and (3) reconstructing the micro-grid, wherein when the load shedding amount is the same and the results meet constraint conditions, the objective function of the micro-grid needs to meet the goal of minimum comprehensive evaluation function value, and is as follows:
Figure BDA0002449579200000121
in the formula: k is a radical of1，k2，k3Weight coefficients for network loss, voltage deviation and power deviation of balance node, and 0 < k1,k2,k3＜1，k1+k2+k3＝1；
Figure BDA0002449579200000122
Respectively, the normalized processing values of the network loss, the voltage deviation and the power deviation of the balance node.
S43: before comprehensive evaluation, the output result is firstly normalized, and the calculation method is as follows:
Figure BDA0002449579200000123
in the formula: l is the system network loss value, L*Is a set network loss reference value; Δ U is the voltage deviation, which is the difference between the highest voltage and the lowest voltage of the node in the whole system, U*Is a system reference voltage; delta PtFor power deviation, it means the deviation of the power of the balance node from the set reference power, Pt *The reference power of the balance node is the average value of the upper limit and the lower limit of the power of the balance node. Wherein,
Figure BDA0002449579200000124
in the formula: pBj、QBj、RBjRespectively, active power, reactive power and resistance corresponding to branch j, wherein QBjThe method is obtained according to the load power factor and the active power; u shapeBjThe voltage at the j end of the branch thereof; r isjFor the branch state after corresponding reconstruction, 1 represents connection, and 0 represents disconnection; j is the number of branches contained in the system.
More specifically, in fig. 2, it is assumed that branch impedance and load data in the microgrid are as shown in table 1. A2.5 MW micro gas turbine MT is installed at a node 1, a 2MW photovoltaic power generation PV1 is installed at a node 3, and a 1MW fan WG1 is installed at a node 6. Wherein, the gas turbine is used as a balance node of the system, and the rest generators and loads are in a PQ mode
TABLE 1 Branch impedance and load data
Figure BDA0002449579200000131
Figure BDA0002449579200000141
Before simulation, setting the active power constraint range of the gas turbine to be 0-2.1 MW; the voltage constraint range of each node is 0.9-1.1; the grid voltage reference value is 12.66kV, the gas turbine is a networking DG, the gas turbine is used as a balance node in load flow calculation, and the rest generators and loads are in a PQ mode. Assuming that the micro-grid needs to be reconstructed and 0.9-1.1MW of active load in the grid needs to be removed, the method is adopted for reconstruction. Wherein table 2 represents the first-level optimization results, table 3 represents the second-level optimization results, and table 5 represents the optimal reconstruction results as follows:
TABLE 2 first order optimization results
Figure BDA0002449579200000142
TABLE 3 second-stage optimization results
Figure BDA0002449579200000143
When the load is cut off to be 0.9MW, the active power of the balance nodes corresponding to the 3 groups of structures is larger than the capacity of the gas turbine, and the balance relation between the load capacity and the generated energy is not satisfied, so that the 3 groups of switch combinations are eliminated. When the next load shedding amount is 1.0MW, the active power of the balance nodes of the corresponding 3 groups of structures is smaller than the capacity of the gas turbine, and the balance relation between the load amount and the generated energy is met; and the minimum voltage and the maximum voltage are both in an allowable range, so that 3 groups of structures with the load shedding amount of 1.0MW are comprehensively evaluated to determine an optimal solution.
The weight of the comprehensive evaluation function (i.e. equation (9)) is selected as k1＝0.8,k2＝0.1,k30.1. Table 4 shows the results of comprehensive evaluation of 1.0MW load shedding amount.
TABLE 4 comprehensive evaluation results for load shedding amount of 1MW
Figure BDA0002449579200000151
TABLE 5 optimal reconstruction results for the method of the invention
Figure BDA0002449579200000152
In order to show the advantages of the method of the present invention, the calculation is performed by comparing with the hierarchical optimization method for directly calculating the load flow, and the optimal solution result can be obtained as shown in table 6.
TABLE 6 optimal reconstruction results for the comparative methods
Figure BDA0002449579200000153
As can be seen from the results in tables 5 and 6, the final switching combinations (i.e., reconstruction schemes) obtained by the two methods are consistent, indicating that the two methods are the same in accuracy. Compared with the reconstruction time of the method and the comparison method, the reconstruction time of the method is 0.073s, the reconstruction time of the comparison method is 6.571s, the time is shortened by about 90 times, and the calculation efficiency is obviously improved.
The above description is only an embodiment of the present invention, but the technical features of the present invention are not limited thereto, and any changes or modifications within the technical field of the present invention by those skilled in the art are covered by the claims of the present invention.

Claims (8)

1. A micro-grid hierarchical optimization reconstruction method based on a BP neural network is characterized by comprising the following steps:
s1: establishing a micro-grid optimization reconstruction model, and establishing a target function and constraint conditions of micro-grid optimization reconstruction;
s2: establishing a BP neural network model, and training and testing the neural network;
s3: by utilizing a hierarchical optimization thought, in the reconstruction of the micro-grid, the optimization process which does not involve load flow calculation is put into the first-stage processing, and the optimization process which involves load flow calculation is taken as the second-stage optimization processing; only the scheme meeting the first-stage optimization condition is subjected to second-stage optimization processing, and a trained BP neural network is used for replacing load flow calculation in the second-stage optimization process;
s4: introducing a comprehensive evaluation method, constructing a comprehensive evaluation function, carrying out comprehensive evaluation on the output result of the BP network, and selecting an optimal reconstruction scheme.
2. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein in the step S1, under a constraint condition, an objective function of the microgrid optimization reconstruction is as follows:
s11: the micro-grid reconstruction needs to meet the goal of minimum load shedding amount, and the objective function is as follows:
Figure FDA0002449579190000011
wherein i ∈ omega is a node set for removing load after reconstruction, and SiRepresenting the load amount corresponding to the node i;
the stable operation of the reconstructed microgrid needs to meet the following constraint conditions:
1) balancing node power constraints:
Ptmin≤Pt≤Ptmax
in the formula: ptActive power adjustable for the balancing node t; ptmaxThe upper limit of the active power can be adjusted for the node t; ptminThe lower limit of the active power can be adjusted for the node t;
2) branch power constraint:
PBj≤PBj max
in the formula: pGxThe power generation power of a micro power source X in the micro power grid is represented, wherein the X represents the number of the micro power sources reserved after reconstruction; pLiLoad active power reserved for the nodes i after the microgrid is reconstructed, wherein N represents the number of the nodes reserved after the reconstruction;
3) and power balance constraint:
Figure FDA0002449579190000021
in the formula: pGxThe power generation power of a micro power source X in the micro power grid is represented, wherein the X represents the number of the micro power sources reserved after reconstruction; pLiLoad active power reserved for the nodes i after the microgrid is reconstructed, wherein N represents the number of the nodes reserved after the reconstruction;
4) micro-power source power generation constraint:
PG min≤PG≤PG max
in the formula: pGThe total generated power of the micro power supply after the micro power grid is reconstructed; pG minThe lower limit of the generated power in the micro-grid; pG maxThe upper limit of the generated power in the micro-grid;
5) node voltage constraint:
Uimin≤Ui≤Uimax
in the formula: u shapeiIs the voltage magnitude of node i; u shapeiminIs the lower voltage limit of node i; u shapeimaxIs the upper voltage limit of node i.
3. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein the BP neural network in the step S2 is constructed as follows:
s21: in order to improve the training precision, the neural network model constructed by the invention comprises 2 hidden layers, and the node number of each hidden layer is determined according to the following formula:
Figure FDA0002449579190000031
in the formula, h is the number of nodes of the hidden layer, z and v are the number of nodes of the input layer and the output layer respectively, and c is an adjusting constant between 1 and 10.
4. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein the training test process of the BP neural network in the step S2 is as follows:
s22: calculating the power flows of the micro-grid corresponding to different switch states by using a Matpower toolkit, wherein the power flows comprise balance node power PtNetwork loss L, node maximum voltage UmaxAnd a minimum voltage Umin(ii) a Collecting the data;
s23: in the collected data, 80% of the data is selected as a training set at will, and a neural network toolbox in Matlab is used for training;
s24: and aiming at the BP neural network obtained by training, introducing the rest 20% of the BP neural network as test set data into the BP neural network, comparing the output result with the data in the test set, and judging that the BP neural network meets the requirements.
5. The microgrid hierarchical optimization reconstruction method based on the BP neural network as claimed in claim 1, wherein the process of applying the hierarchical optimization idea to the microgrid reconstruction in the step S3 is as follows:
s31: in the first-stage optimization process, processing an objective function and constraint which do not involve any load flow calculation; considering the balance between the generated energy and the load quantity, performing integer programming to obtain a load switch state combination meeting the balance condition, and sequencing according to the ascending order of the power supply load shedding quantity to obtain a switch combination solution set D;
s32: in the second-stage optimization process, processing an objective function and constraint related to load flow calculation, substituting the switch combination solution set D obtained by the first-stage optimization into a BP neural network one by one for prediction, and obtaining a prediction result;
s33: judging whether the prediction result meets the relevant constraint of the stable operation of the micro-grid or not for the load shedding amount; if only one group of switch combination is satisfied, the switch combination is the optimal solution of the micro-grid reconstruction; on the contrary, if the output results of a plurality of groups of switch combinations meet the constraint, selecting the optimal solution of the micro-grid reconstruction according to the comprehensive evaluation method; and if all the combinations are not satisfied, selecting the next load shedding amount to repeat the steps until the optimal solution is found.
6. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein the comprehensive evaluation method in the step S4 is as follows:
s41: if the output results of a plurality of groups of switch combinations meet the constraint condition in the same load cutting amount, a plurality of groups of corresponding network loss, voltage deviation and power deviation outputs can be obtained; setting reference values of network loss, voltage deviation and power deviation, normalizing the output results of each group of switches, and selecting corresponding weight coefficients k according to the preference of a decision makeriAnd selecting the switch combination with the minimum comprehensive evaluation function value as the optimal solution by utilizing the comprehensive evaluation function.
7. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein the comprehensive evaluation function in step S4 is constructed as follows:
s42: and (3) reconstructing the micro-grid, wherein when the load shedding amount is the same and the results meet constraint conditions, the objective function of the micro-grid needs to meet the goal of minimum comprehensive evaluation function value, and is as follows:
Figure FDA0002449579190000041
in the formula: k is a radical of1，k2，k3Weight coefficients for network loss, voltage deviation and power deviation of balance node, and 0 < k1,k2,k3＜1，k1+k2+k3＝1；
Figure FDA0002449579190000042
Respectively normalizing values of network loss, voltage deviation and power deviation of a balance node;
s43: before comprehensive evaluation, the output result is firstly normalized, and the calculation method is as follows:
Figure FDA0002449579190000043
in the formula: l is the system network loss value, L*Is a set network loss reference value; Δ U is the voltage deviation, which is the difference between the highest voltage and the lowest voltage of the node in the whole system, U*Is a system reference voltage; delta PtFor power deviation, it means the deviation of the power of the balance node from the set reference power, Pt *The reference power of the balance node is the average value of the upper and lower limits of the power of the balance node; wherein,
Figure FDA0002449579190000051
in the formula: pBj、QBj、RBjRespectively, active power, reactive power and resistance corresponding to branch j, wherein，QBjThe method is obtained according to the load power factor and the active power; u shapeBjThe voltage at the j end of the branch thereof; r isjFor the branch state after corresponding reconstruction, 1 represents connection, and 0 represents disconnection; j is the number of branches contained in the system.
8. The microgrid hierarchical optimization reconstruction method based on a BP neural network as claimed in claim 1, wherein in the step S3, an exception handling process is as follows:
s34: if the output results of all switch combinations in the switch state set D do not meet the constraint condition, indicating that the fault reconstruction is abnormal, and entering an abnormal processing mechanism; in all the switch combinations which do not satisfy the constraint condition, the node voltage deviation and the balanced node power threshold value of each switch combination are comprehensively calculated to be more limited Y, and the method is as follows:
Figure FDA0002449579190000052
in the formula, w1，w2Is a weight coefficient of 0<w1，w2<1，w1+w21 is ═ 1; i is the minimum voltage node, dUiminIs the lower out-of-limit absolute value of the minimum voltage, UiminIs the lower limit value of the node voltage; j is the maximum voltage node, dUjmaxIs the upper out-of-limit absolute value of the maximum voltage, UjmaxIs a node voltage upper limit value; t is a balance node, dPtminTo balance the absolute value of the off-limit power at the node, PtminTo balance node power lower limit, dPtmaxTo balance the absolute value of the out-of-limit power on the node, PtmaxBalancing the node power upper limit value; and selecting the switch combination with the minimum comprehensive limit Y as the optimal reconstruction output solution.
CN202010288722.7A 2020-04-14 2020-04-14 BP neural network-based micro-grid hierarchical optimization reconstruction method Pending CN111680815A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010288722.7A CN111680815A (en) 2020-04-14 2020-04-14 BP neural network-based micro-grid hierarchical optimization reconstruction method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010288722.7A CN111680815A (en) 2020-04-14 2020-04-14 BP neural network-based micro-grid hierarchical optimization reconstruction method

Publications (1)

Publication Number Publication Date
CN111680815A true CN111680815A (en) 2020-09-18

Family

ID=72451631

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202010288722.7A Pending CN111680815A (en) 2020-04-14 2020-04-14 BP neural network-based micro-grid hierarchical optimization reconstruction method

Country Status (1)

Country Link
CN (1) CN111680815A (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113629780A (en) * 2021-08-11 2021-11-09 山东大学 Microgrid power converter control method, system, storage medium and device

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106058862A (en) * 2016-07-11 2016-10-26 南昌大学 Distribution network rapid reconstruction method taking voltage stability into consideration
CN106655174A (en) * 2017-01-03 2017-05-10 昆明理工大学 Comprehensive reconstruction optimization method for power distribution network

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106058862A (en) * 2016-07-11 2016-10-26 南昌大学 Distribution network rapid reconstruction method taking voltage stability into consideration
CN106655174A (en) * 2017-01-03 2017-05-10 昆明理工大学 Comprehensive reconstruction optimization method for power distribution network

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
张艳伟等: "岸桥调度的三维空间建模方法研究", 《武汉理工大学学报（交通科学与工程版）》 *
李咸善等: "微电网分级优化故障重构", 《中国电机工程学报》 *
许锐等: "基于BP神经网络的电力系统潮流计算", 《电子世界》 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113629780A (en) * 2021-08-11 2021-11-09 山东大学 Microgrid power converter control method, system, storage medium and device
CN113629780B (en) * 2021-08-11 2023-03-24 山东大学 Microgrid power converter control method, system, storage medium and device

Similar Documents

Publication Publication Date Title
CN103310065B (en) Meter and the intelligent network distribution reconstructing method of distributed power generation and energy-storage units
CN107546743B (en) Distributed power flow optimization method for radial power distribution network
CN110970906B (en) Method and system for coordinated control of reactive voltage of wind power plant
CN102868161A (en) Optimization method of network variable structure with distributed type power supply distribution system
CN107069814A (en) Fuzzy opportunity constraint planning method and system for distribution network distributed power capacity distribution
CN113241803B (en) Energy storage scheduling method based on new energy consumption and computer medium
CN109818347B (en) Assessment method for wind power consumption capability of electric power system
CN113378100B (en) Power distribution network source network load storage collaborative optimization scheduling model and method considering carbon emission
CN106058858A (en) Power distribution network optimization method and device
CN114726009B (en) Wind power plant group reactive power hierarchical optimization control method and system considering power prediction
CN113962159A (en) Method for evaluating maximum photovoltaic access capacity of power distribution network based on reasonable light abandonment
CN106229995B (en) Based on the backup power source shunt reactor parameter optimization method under the Anti-Typhoon operational mode of wind power plant
CN114069708A (en) Uncertainty-considered dispersed optical storage multi-level multi-objective optimization access method
CN110783913B (en) Group-based optimal power grid topology online optimization method considering expected accident set
CN107947183A (en) The power distribution network adaptive optimization method containing three end SNOP based on differential evolution
CN118333219A (en) Double-layer collaborative optimization configuration method and device considering wind-solar storage charging access
CN114491886A (en) General modeling method and device for active power distribution network containing multi-type distributed new energy
CN111680815A (en) BP neural network-based micro-grid hierarchical optimization reconstruction method
CN112632748A (en) Power system stability risk optimization method, system and storage medium
jing Hu et al. Capacity optimization of wind/PV/storage power system based on simulated annealing-particle swarm optimization
CN110034585A (en) A kind of quick planing method of distributed generation resource
CN112653185B (en) Efficiency evaluation method and system of distributed renewable energy power generation system
CN115549118A (en) Output smoothness index prediction method for reactive power compensation equipment of power distribution network
CN112182952B (en) Multi-objective optimal scheduling method for improving elasticity of power system
Fathi et al. Multi-Objective optimization of islanded microgrids

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20200918

RJ01 Rejection of invention patent application after publication