Acquisition method of multi-target reactive power optimization model containing distributed power supply

Abstract

The invention discloses a method for acquiring a multi-target reactive power optimization model containing a distributed power supply. The method comprises the following steps of: 1, establishing a random output probability model containing wind power and photovoltaic power; 2, converting the random output probability model into a probabilistic load flow calculation model by using a random response surface method RSM; and step 3, establishing a multi-target reactive power optimization model containing the distributed power supply based on the probabilistic load flow calculation model. The objective of the invention is to solve the technical problems that an existing reactive power optimization model does not consider uncertainty caused by grid connection of distributed power supplies, reactive power flow in a network is changed, reactive power of a certain node is insufficient or excessive, and thus unstable operation of a power grid is possibly caused, and even a power failure event is caused. The invention provides a multi-target reactive power optimization technology containing the distributed power supply.

Description

Technical Field

[0001] The present invention relates to the field of new energy technology, in particular to reactive power optimization technology for a power grid containing new energy, and specifically to a multi-objective reactive power optimization model containing distributed power sources and an acquisition method thereof. Background Art

[0002] At present, the world's energy crisis and environmental problems are becoming increasingly serious. In order to achieve the strategic goal of sustainable development, sustainable energy supply is essential. Therefore, the development and utilization of new energy has become an important strategy for energy development in countries around the world. The output of distributed power sources is highly random. After being connected to the power grid, it will change the power flow distribution of the original system, have an adverse effect on the network loss and node voltage of the system, and even affect the safety and stability of the system, making the reactive power optimization problem of the distribution network more complicated. Therefore, it is necessary to study the reactive power optimization of the power system containing distributed power sources.

[0003] The patent document with application publication number CN104104096A discloses a reactive power optimization method and system for a power system. The method determines the control variables of the reactive power optimization mathematical model by establishing a reactive power optimization mathematical model including a reactive power optimization objective function and reactive power optimization constraints, and solves the reactive power optimization mathematical model using a genetic algorithm. The patent document with application publication number CN104600720A discloses a method for voltage reactive power optimization control of a regional power grid. The method changes the relationship between multiple objectives in reactive power optimization by changing the weight of the scoring basis, thereby enhancing the maintainability and scalability of the online reactive power optimization control system of the regional power grid, and enhancing the maintainability and scalability of the system. The invention proposes a reactive power optimization control system architecture based on multi-agent technology, which not only avoids a large number of redundant rules in the suboptimal control of the expert system, but also simply changes the relationship between multiple objectives in reactive power optimization by changing the weight of the scoring basis, greatly enhancing the maintainability and scalability of the system.

[0004] In summary, the traditional reactive power optimization model does not take into account the uncertainty caused by the grid connection of distributed power sources, which may cause changes in the reactive power flow in the network, resulting in insufficient or excessive reactive power at a certain node, which may lead to unstable grid operation and even cause power outages. Summary of the invention

[0005] The purpose of the present invention is to solve the technical problem that the existing reactive power optimization model does not take into account the uncertainty brought about by the grid connection of distributed power sources, which causes the reactive power flow in the network to change, resulting in insufficient or excessive reactive power at a certain node, which may lead to unstable grid operation and even cause power outages. A multi-objective reactive power optimization model containing distributed power sources and an acquisition method are proposed.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A method for obtaining a multi-objective reactive power optimization model containing distributed power sources comprises the following steps:

[0008] Step 1: Establish a random output probability model including wind power and photovoltaic power;

[0009] Step 2: Use the random response surface method (RSM) to convert the random output probability model into a probabilistic power flow calculation model;

[0010] Step 3: Establish a multi-objective reactive power optimization model containing distributed generation based on the probabilistic power flow calculation model.

[0011] In step 1: establish a random output probability model including wind power and photovoltaic power; specifically:

[0012] 1) The probability model of wind power output is:

[0013] Assuming that the change in wind speed follows a normal distribution, its probability density function is:

[0014]

[0015] Where: σ is the standard deviation; μ is the expected value; v is the wind speed.

[0016] The active power output of the fan is shown in formula (2):

[0017]

[0018] Where: P ra is the rated power of the fan; v in 、v out 、v ra They are the cut-in, cut-out and rated wind speed of the fan respectively;

[0019] 2) The probability model of photovoltaic power generation output is:

[0020] Photovoltaic output is mainly related to light intensity H, solar panel area S, and photoelectric conversion efficiency λ. Photovoltaic output can be expressed as the product of H, S, and λ.

[0021] Assuming that the change of light intensity H follows Beta distribution, its probability density function is:

[0022]

[0023] Where: H max is the maximum light intensity; β and γ are the shape parameter and size parameter of Beta distribution respectively.

[0024] In step 2: the steps for model transformation using random response surface methodology are as follows:

[0025] Step 2-1: Convert the input random variables into standard random variables. That is, convert the normal distribution (μ, σ) describing the wind turbine output into μ+σξ, and convert the Beta distribution (α, β) describing the photovoltaic output into ξ is the standard (0,1) normal distribution;

[0026] Step 2-2: The two random variables of wind power generation output and photovoltaic power generation output are used as the system input uncertain parameters x1 and x2, and the solutions of each item of power flow calculation are used as the system output responses Y1 and Y2. The relationship between the response and the input parameters is expressed as a "black box" model Y = h(x);

[0027] Step 2-3: Express each unknown quantity in the power flow calculation in the form of a third-order Hermite polynomial. First, multiply the unknown coefficient of the ith Hermite polynomial by the corresponding n-order Hermite polynomial represented by n random variables, and then add these products.

[0028] Step 2-4: Determine the collocation values ​​of random variables. Generate collocation values ​​of the input uncertain parameters wind turbine power and photovoltaic power according to the collocation method of random response surface method;

[0029] Step 2-5: Calculate the response value at the distribution point value. In the deterministic model, call the conventional power flow program to calculate the solutions of various quantities to be determined in the power flow calculation at the wind turbine and photovoltaic power distribution point values;

[0030] Step 2-6: Calculate the undetermined coefficients of the Hermite polynomial. Substitute the point values ​​obtained in step 2-4 and the solutions of the corresponding flow calculation quantities into the Hermite polynomial obtained in step 2-2 to form a linear equation system. Solving the linear equation system will give the values ​​of the undetermined coefficients.

[0031] Step 2-7: Calculate the statistical characteristic value or probability density function of the response. Substituting the unknown coefficients obtained in the previous step into the Hermite polynomial, the probability distribution function of each unknown quantity in the power flow calculation can be obtained.

[0032] In step 3, a multi-objective reactive power optimization model containing distributed generation is established based on the probabilistic power flow calculation model. The model includes an objective function, equality constraints and inequality constraints with the optimization objectives of minimizing system network loss and voltage deviation.

[0033] A multi-objective reactive power optimization model is established with the optimization objectives of minimizing the system active network loss and node voltage deviation. Specifically:

[0034] 1) Objective function

[0035]

[0036] Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of over-limit voltage nodes; α is the node voltage penalty function coefficient; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of the node voltage respectively.

[0037] When V i <V i,min When △V i It can be expressed as V i -V i,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max .

[0038] 2) Equality constraints

[0039]

[0040] Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss respectively.

[0041] 3) Inequality constraints

[0042]

[0043] Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

[0044] The method also includes step 4, using the multi-objective differential grey wolf algorithm MODGWO to process the model and obtain the final reactive power optimization result; specifically:

[0045] Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters;

[0046] Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population;

[0047] Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution;

[0048] Step 4-4: Determine whether the current number of iterations k meets the iteration requirement. If the number of iterations is not reached, execute step 4-5; if the number of iterations is reached, output the optimal Pareto solution;

[0049] Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population;

[0050] Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population. Same as step 4-3, select the alpha wolf in the new population and return to step 4-4.

[0051] A multi-objective reactive power optimization model containing distributed power sources is established based on a probabilistic power flow calculation model. The model includes an objective function, equality constraints and inequality constraints with the minimum system network loss and voltage deviation as the optimization objectives.

[0052] A multi-objective reactive power optimization model is established with the optimization objectives of minimizing the system active network loss and node voltage deviation. Specifically:

[0053] 1) Objective function

[0054]

[0055] Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of over-limit voltage nodes; α is the node voltage penalty function coefficient; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of the node voltage respectively.

[0056] When V i <V i,min When △V i It can be expressed as V i -V i,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max .

[0057] 2) Equality constraints

[0058]

[0059] Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss respectively.

[0060] 3) Inequality constraints

[0061]

[0062] Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

[0063] A multi-objective differential grey wolf algorithm comprises the following steps:

[0064] Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters;

[0065] Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population;

[0066] Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution;

[0067] Step 4-4: Determine whether the current number of iterations k meets the iteration requirement. If the number of iterations is not reached, execute step 4-5; if the number of iterations is reached, output the optimal Pareto solution;

[0068] Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population;

[0069] Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population. Same as step 4-3, select the alpha wolf in the new population and return to step 4-4.

[0070] This method is used to calculate the multi-objective reactive power optimization model and obtain the final reactive power optimization result.

[0071] Compared with the prior art, the present invention has the following technical effects:

[0072] 1) In order to solve the uncertainty problem caused by the grid connection of distributed power sources, the present invention fully considers the impact of the randomness of the output of various distributed power sources on the distribution network, and establishes a multi-objective reactive power optimization model containing distributed power sources. After the reactive power optimization of the distribution network, the system network loss and voltage offset are reduced, and the overall system performance is optimized;

[0073] 2) In the process of solving the multi-objective reactive power optimization model, the present invention proposes a multi-objective differential gray wolf algorithm. This algorithm integrates the mutation and crossover in the differential algorithm and the multi-objective processing method, which can maintain the advantage of high convergence speed of the original gray wolf algorithm, while solving the problem of easily falling into local optimal solutions and multi-objective balance. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] The present invention will be further described below in conjunction with the accompanying drawings and embodiments:

[0075] Figure 1 is a flow chart of the method of the present invention;

[0076] Figure 2 It is a flow chart of the random response surface method RSM proposed by the present invention;

[0077] Figure 3 It is a flow chart of the multi-objective differential grey wolf algorithm MODGWO proposed by the present invention;

[0078] Figure 4 It is a schematic diagram of the IEEE 39-node system in an example of the present invention. DETAILED DESCRIPTION

[0079] like Figure 1 As shown, a method for solving a multi-objective reactive power optimization model containing distributed power sources includes the following steps:

[0080] Step 1: Establish a random output probability model including wind power and photovoltaic power;

[0081] Step 2: Use the random response surface method (RSM) to convert the random output probability model into a probabilistic power flow calculation model;

[0082] Step 3: Establish a multi-objective reactive power optimization model including distributed generation based on the probabilistic power flow calculation model;

[0083] Step 4: Use the multi-objective differential grey wolf algorithm MODGWO to process the model and obtain the final reactive power optimization result.

[0084] In step 1: establish a random output probability model including wind power and photovoltaic power.

[0085] 1) Wind power output probability model

[0086] Assuming that the change in wind speed follows a normal distribution, its probability density function is:

[0087]

[0088] Where: σ is the standard deviation; μ is the expected value; v is the wind speed.

[0089] The active power output of the fan is shown in formula (2):

[0090]

[0091] Where: P ra is the rated power of the fan; v in 、v out 、v ra They are the cut-in, cut-out and rated wind speed of the fan respectively;

[0092] 2) Photovoltaic power generation output probability model

[0093] Photovoltaic output is mainly related to light intensity H, solar panel area S, and photoelectric conversion efficiency λ. Photovoltaic output can be expressed as the product of H, S, and λ.

[0094] Assuming that the change of light intensity H follows Beta distribution, its probability density function is:

[0095]

[0096] Where: H max is the maximum light intensity; β and γ are the shape parameter and size parameter of Beta distribution respectively.

[0097] In step 2: use the random response surface method (RSM) to convert the random output probability model into a probabilistic power flow calculation model. The steps are as follows:

[0098] Step 2-1: Convert the input random variables into standard random variables. That is, convert the normal distribution (μ, σ) describing the wind turbine output into μ+σξ, and convert the Beta distribution (α, β) describing the photovoltaic output into ξ is the standard (0,1) normal distribution;

[0099] Step 2-2: The two random variables of wind power generation output and photovoltaic power generation output are used as the system input uncertain parameters x1 and x2, and the solutions of each item of power flow calculation are used as the system output responses Y1 and Y2. The relationship between the response and the input parameters is expressed as a "black box" model Y = h(x);

[0100] Step 2-3: Express each unknown quantity in the power flow calculation in the form of a third-order Hermite polynomial. First, multiply the unknown coefficient of the ith Hermite polynomial by the corresponding n-order Hermite polynomial represented by n random variables, and then add these products.

[0101] Step 2-4: Determine the collocation values ​​of random variables. Generate collocation values ​​of the input uncertain parameters wind turbine power and photovoltaic power according to the collocation method of random response surface method;

[0102] Step 2-5: Calculate the response value at the distribution point value. In the deterministic model, call the conventional power flow program to calculate the solutions of various quantities to be determined in the power flow calculation at the wind turbine and photovoltaic power distribution point values;

[0103] Step 2-6: Calculate the undetermined coefficients of the Hermite polynomial. Substitute the point values ​​obtained in step 2-4 and the solutions of the corresponding flow calculation quantities into the Hermite polynomial obtained in step 2-2 to form a linear equation system. Solving the linear equation system will give the values ​​of the undetermined coefficients.

[0104] Step 2-7: Calculate the statistical characteristic value or probability density function of the response. Substituting the unknown coefficients obtained in the previous step into the Hermite polynomial, the probability distribution function of each unknown quantity in the power flow calculation can be obtained.

[0105] In step 3: a multi-objective reactive power optimization model including distributed generation is established based on the probabilistic power flow calculation model. The model includes an objective function, equality constraints and inequality constraints with the optimization objectives of minimizing system network loss and voltage deviation.

[0106] A multi-objective reactive power optimization model is established with the optimization objectives of minimizing system active network losses and node voltage deviation.

[0107] 1) Objective function

[0108]

[0109] Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of over-limit voltage nodes; α is the node voltage penalty function coefficient; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of the node voltage respectively.

[0110] When V i <V i,min When △V i It can be expressed as V i -Vi,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max .

[0111] 2) Equality constraints

[0112]

[0113] Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss respectively.

[0114] 3) Inequality constraints

[0115]

[0116] Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

[0117] In step 4: the multi-objective differential grey wolf algorithm MODGWO is used to process the model and obtain the final reactive power optimization result.

[0118] Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters;

[0119] Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population;

[0120] Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution;

[0121] Step 4-4: Determine whether the current number of iterations k meets the iteration requirement. If the number of iterations is not reached, execute step 4-5; if the number of iterations is reached, output the optimal Pareto solution;

[0122] Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population;

[0123] Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population. Same as step 4-3, select the alpha wolf in the new population and return to step 4-4.

[0124] The present invention also includes a multi-objective reactive power optimization model containing distributed power sources, which is characterized in that a multi-objective reactive power optimization model containing distributed power sources is established based on a probabilistic power flow calculation model, and the model includes an objective function, equality constraints and inequality constraints with minimization of system network loss and voltage deviation as optimization objectives.

[0125] A multi-objective reactive power optimization model is established with the optimization objectives of minimizing the system active network loss and node voltage deviation. Specifically:

[0126] 1) Objective function

[0127]

[0128] Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of over-limit voltage nodes; α is the node voltage penalty function coefficient; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of the node voltage respectively.

[0129] When V i <V i,min When △V i It can be expressed as V i -V i,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max .

[0130] 2) Equality constraints

[0131]

[0132] Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss respectively.

[0133] 3) Inequality constraints

[0134]

[0135] Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

[0136] The model fully considers the impact of the randomness of the output of various distributed power sources on the distribution network. After the model optimizes the reactive power of the distribution network, the system network loss and voltage offset are reduced, and the overall performance of the system is optimized.

[0137] The present invention also includes a multi-objective differential grey wolf algorithm, comprising the following steps:

[0138] Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters;

[0139] Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population;

[0140] Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution;

[0141] Step 4-4: Determine whether the current number of iterations k meets the iteration requirement. If the number of iterations is not reached, execute step 4-5; if the number of iterations is reached, output the optimal Pareto solution;

[0142] Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population;

[0143] Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population. Same as step 4-3, select the alpha wolf in the new population and return to step 4-4.

[0144] Example:

[0145] In order to verify the effectiveness of the present invention, the MODGWO algorithm is simulated in the IEEE 39-node model, and wind turbines and photovoltaic generators are added to the system. Figure 4 As shown, synchronous generator sets G3 and G5 are replaced with wind generator sets, and synchronous generator set G7 is replaced with photovoltaic generator sets. The IEEE 39-node system reference voltage is 12.66kV, the reference power is 100MVA, and the total load is 3.8+j2.6MVA. 10 groups of switchable capacitors with a reactive capacity of 100kVar are connected at nodes 4, 6 and 32 respectively, and the penalty factor is 10000. The number of wolf packs in the MODGWO algorithm is 100, the number of iterations is 100, and the node voltage deviation is within 5%. The wind power generation parameters and photovoltaic power generation parameters are shown in Table 1.

[0146] In order to verify the effectiveness of the optimization model constructed by the present invention, simulation tests were performed on the IEEE 39-node system without DG and with DG. The control variable parameters and optimization results before and after the system optimization are shown in Table 2 and Table 3, respectively. It can be seen from Table 3 that compared with before the DG was connected to the grid, the system active network loss dropped to 131.56kW, a decrease of 18.43%, and the voltage offset dropped to 1.1365pu, a decrease of 12.47%. It can be seen that the access of DG is conducive to the improvement of system safety and economy. After the MODGWO algorithm was used for reactive power optimization, the system active network loss and voltage offset further decreased, by 55.82% and 51.08%, respectively. At the same time, it can be seen from Table 2 that compared with before the optimization, the input capacity of the discrete control equipment was also significantly reduced, saving the equipment investment of reactive power compensation required for reactive power optimization of the distribution network. In summary, the optimization model and solution method constructed by the present invention can achieve the purpose of reducing system network loss and voltage offset, thereby ensuring the safety of system operation.

[0147] Table 1

[0148]

[0149] Table 2

[0150]

[0151] Table 3

[0152]

[0153] The method provided by the present invention effectively reduces the network loss and voltage deviation of the system by optimizing the reactive power distribution of the distribution network, thereby improving the overall performance. In addition, the present invention proposes a multi-objective differential gray wolf algorithm, which combines the mutation and crossover strategies in the differential algorithm with the multi-objective processing technology, not only maintaining the fast convergence characteristics of the gray wolf algorithm, but also avoiding falling into the local optimum and balancing the multi-objective optimization requirements.

Claims

1. A method for obtaining a multi-objective reactive power optimization model containing distributed power sources, characterized in that: The following steps are involved: Step 1: Establish a random output probability model including wind power and photovoltaic power; Step 2: Use the random response surface method (RSM) to convert the random output probability model into a probabilistic power flow calculation model; Step 3: Establish a multi-objective reactive power optimization model containing distributed generation based on the probabilistic power flow calculation model.

2. The method according to claim 1, characterized in that In step 1: establish a random output probability model including wind power and photovoltaic power; specifically: 1) The probability model of wind power output is: Assuming that the change in wind speed follows a normal distribution, its probability density function is: Where: σ is the standard deviation; μ is the expected value; v is the wind speed; The active power output of the fan is shown in formula (2): Where: P ra is the rated power of the fan; v in 、v out 、v ra They are the cut-in, cut-out and rated wind speed of the fan respectively; 2) The probability model of photovoltaic power generation output is: Photovoltaic output is mainly related to light intensity H, solar panel area S, and photoelectric conversion efficiency λ. Photovoltaic output can be expressed as the product of H, S, and λ. Assuming that the change of light intensity H follows Beta distribution, its probability density function is: Where: H max is the maximum light intensity; β and γ are the shape parameter and size parameter of Beta distribution respectively.

3. The method according to claim 1, characterized in that In step 2: the steps for model transformation using random response surface methodology are as follows: Step 2-1: Convert the input random variables into standard random variables; that is, convert the normal distribution (μ, σ) describing the wind turbine output into μ+σξ, and convert the Beta distribution (α, β) describing the photovoltaic output into ξ is the standard (0,1) normal distribution; Step 2-2: The two random variables of wind power generation output and photovoltaic power generation output are used as the system input uncertain parameters x1 and x2, and the solutions of each item of power flow calculation are used as the system output responses Y1 and Y2. The relationship between the response and the input parameters is expressed as a "black box" model Y = h(x); Step 2-3: Express each unknown quantity in the power flow calculation in the form of a third-order Hermite polynomial. First, multiply the unknown coefficient of the ith Hermite polynomial by the corresponding n-order Hermite polynomial represented by n random variables, and then add these products. Step 2-4: Determine the collocation value of the random variable; generate the collocation value of the input uncertain parameters wind turbine power and photovoltaic power according to the collocation method of the random response surface method; Step 2-5: Calculate the response value at the distribution point value; call the conventional power flow program in the deterministic model to calculate the solutions of various quantities to be determined in the power flow calculation at the wind turbine and photovoltaic power distribution point values; Step 2-6: Calculate the undetermined coefficients of the Hermite polynomial; Substitute the solution of the collocation point value and the corresponding flow calculation variables obtained in step 2-4 into the Hermite polynomial obtained in step 2-2 to form a linear equation system. Solving the linear equation system will obtain the values ​​of the undetermined coefficients; Step 2-7: Calculate the statistical characteristic value or probability density function of the response; substitute the unknown coefficients obtained in the previous step into the Hermite polynomial to obtain the probability distribution function of each unknown quantity in the power flow calculation.

4. The method according to claim 1, characterized in that In step 3, a multi-objective reactive power optimization model containing distributed generation is established based on the probabilistic power flow calculation model. The model includes an objective function, equality constraints and inequality constraints with the optimization objectives of minimizing system network loss and voltage deviation.

5. The method according to claim 4, characterized in that A multi-objective reactive power optimization model is established with the optimization objectives of minimizing the system active network loss and node voltage deviation. Specifically: 1) Objective function Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of nodes with over-limit voltage; α is the penalty function coefficient of node voltage; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of node voltage respectively; When V i <V i,min When △V i It can be expressed as V i -V i,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max ; 2) Equality constraints Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss; 3) Inequality constraints Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

6. The method according to any one of claims 1 to 5, characterized in that: The method also includes step 4, using the multi-objective differential grey wolf algorithm MODGWO to process the model and obtain the final reactive power optimization result; specifically: Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters; Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population; Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution; Step 4-4: Determine whether the current number of iterations k meets the iteration requirement; if not, execute step 4-5; if it has reached the number of iterations, output the optimal Pareto solution; Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population; Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population; the same operation as step 4-3, select the alpha wolf in the new population and return to step 4-4.

7. A multi-objective reactive power optimization model containing distributed power sources, characterized in that: A multi-objective reactive power optimization model containing distributed generation is established based on the probabilistic power flow calculation model. The model includes objective functions, equality constraints and inequality constraints with the optimization objectives of minimizing system network loss and voltage deviation.

8. The optimization model according to claim 7, characterized in that: A multi-objective reactive power optimization model is established with the optimization objectives of minimizing the system active network loss and node voltage deviation. Specifically: 1) Objective function Where: P loss is the system active network loss; Ω B is the node set in the distribution network; Ω V is the set of over-limit voltage nodes; α is the node voltage penalty function coefficient; △V i is the voltage over-limit penalty function; G ij is the conductance between node i and node j; V i 、V j are the effective values ​​of the voltage at node i and node j respectively; θ ij represents the voltage phase angle between nodes i and j; B is the total number of nodes in the distribution network; V i.max and V i.min are the upper and lower limits of node voltage respectively; When V i <V i,min When △V i It can be expressed as V i -V i,min ; When V i,min <V i <V i,max When △V i is 0; when V i >V i,max When △V i It can be expressed as V i -V i,max ; 2) Equality constraints Where: P G , Q G are respectively the active and reactive power injected into the upper power grid; Ω X is the node set with wind turbines and photovoltaics installed; Ω S is the node set with reactive power compensation devices installed; P X,j , P l,j , P loss,j are the active power of wind power and photovoltaic power injected at node j, the active power required by the load, and the active power loss; Q X,j , Q S,j , Q l,j , Q loss,j are the reactive power of wind power, photovoltaic power, and reactive power device injected at node j, the reactive power required by the load, and the reactive power loss; 3) Inequality constraints Where: I min ,I max are the lower and upper limits of the current respectively; V min 、V max are the lower and upper limits of voltage respectively; I j 、V j are the current and voltage at node j respectively; P X,min , Q X,min , P X,max , Q X,max They are the lower and upper limits of active and reactive output of wind turbines and photovoltaics respectively.

9. A multi-objective differential grey wolf algorithm, characterized in that: The following steps are involved: Step 4-1: Input wind turbine parameters, photovoltaic parameters, population size, number of iterations and other related parameters; Step 4-2: Initialize the parent gray wolf population, calculate the power flow through the random response surface method, and obtain the corresponding objective function values ​​of all solutions in the parent population; Step 4-3: Perform a fast non-dominated sort on the parent population according to the objective function value, select the solution with a sorting level of 1 and store it in the external solution set, and select the leader of the parent population in the external solution set according to the fuzzy membership function of the solution; Step 4-4: Determine whether the current number of iterations k meets the iteration requirement; if not, execute step 4-5; if it has reached the number of iterations, output the optimal Pareto solution; Step 4-5: Update the position of the gray wolf, perform mutation and crossover processing, generate the offspring population and calculate the objective function of the offspring population; Step 4-6: Mix the updated parent population and child population, sort them from large to small based on the crowding distance, and select the previously specified population size solution as the new population; the same operation as step 4-3, select the alpha wolf in the new population and return to step 4-4.

10. The model according to claim 9, characterized in that This method is used to calculate the multi-objective reactive power optimization model and obtain the final reactive power optimization result.

Images