Power distribution network optimization method and system considering distributed generation output and load demand time variation

Abstract

The invention provides a power distribution network optimization method and system considering distributed generation output and load demand time varying, and relates to the technical field of power grids, and the method comprises the following steps: S1, initializing system parameters; s2, collecting distributed photovoltaic output and load requirements; s3, constructing a load state estimation model, and obtaining a measurement missing point load based on the load state estimation model; s4, performing power distribution network reconstruction optimization by using the optimization function; and S5, periodically monitoring node voltage deviation, and when the deviation is greater than a given threshold value, starting network reconstruction of the power distribution network. According to the method, random changes of the load and the generated output can be considered at the same time, and the power distribution network is reconstructed and optimized, so that the voltage level and power loss of the power distribution network are improved. Meanwhile, due to the introduction of the state estimation method, the method is also suitable for the reconstruction problem caused by insufficient coverage of the measurement meter of the power distribution network in the remote area.

Description

Technical Field

[0001] The present invention relates to the technical field of power grids, and in particular to a distribution network optimization method and system that considers time-varying distributed power generation output and load demand. Background Art

[0002] As the main body of the power system, the distribution network directly involves end users and plays a vital role. The network reconstruction of the distribution network changes the topology of the power grid by controlling the switch state to achieve related goals, such as minimizing active power loss or eliminating voltage deviations. It is necessary to consider the complexity of the switching operation and ensure the quality and reliability of power supply, which itself presents certain challenges. At the same time, the growing load demand and the introduction of new technologies such as distributed generation and nonlinear loads have further increased the risk of power outages for users. Therefore, in order to serve the construction of new power systems and improve the operation quality and efficiency of distribution networks, it is necessary to conduct periodic network reconstruction analysis of distribution networks to ensure that they meet the high-quality and uninterrupted power demand of end users.

[0003] Although the time-varying nature of power generation output and load demand is an important factor leading to power outages in distribution networks, the randomness of power generation output and load demand has not received enough attention in previous studies of distribution network optimization strategies. The main goal of existing distribution network optimization algorithms is to minimize the power loss of distribution lines and simply assume that the load is static or increases the load at a certain proportional coefficient, resulting in the optimization results being only suitable for a certain section, and not suitable for the current scenario where the distributed photovoltaic output changes dramatically.

[0004] Since existing studies lack consideration of the time-varying nature of power generation load and new challenging operating conditions, the main goal of the present invention is to bridge the gap in the existing power grid reconstruction strategy of insufficient consideration of the dynamic changes of power generation load, thereby forming a more comprehensive and adaptive distribution network reconstruction strategy. Summary of the invention

[0005] In order to solve the above-mentioned deficiencies of the prior art, the present invention proposes a distribution network optimization method and system that considers the time-varying distributed power generation output and load demand. The method is based on a differential evolution algorithm and can simultaneously consider the random changes of load and power generation output, and reconstruct and optimize the distribution network, thereby improving the voltage level and power loss of the distribution network. At the same time, the introduction of the state estimation method makes the method also applicable to the reconstruction problem of distribution networks in remote areas caused by insufficient coverage of measurement meters.

[0006] Specifically, in a first aspect, the present invention provides a distribution network optimization method considering the time-varying distributed power generation output and load demand, which comprises the following steps:

[0007] S1, system parameter initialization;

[0008] S2, collect distributed photovoltaic output and load demand;

[0009] S3. Construct a load state estimation model, obtain the load of the measurement missing point based on the load state estimation model, and obtain the voltage and current of all nodes. The calculation formula is as follows:

[0010]

[0011] I u =V u Y u

[0012] Among them, I u To measure the missing node branch current value, V u To measure the missing node voltage estimate, V k To measure the normal node voltage value, I k To measure the normal node branch current value, Y k , Y ku , Y u are the elements in the system admittance matrix, where:

[0013] S4. Use the optimization function to optimize the distribution network reconstruction. The optimization function is as follows:

[0014] X p :={T p / X p ,if fit(T p )<fit(X p )}

[0015] Among them, fit( ) is the fitness function, X p is the initial population of disconnected lines in the distribution network, T p is the subpopulation after the gene combination of the disconnected lines in the distribution network; X p : Disconnect line populations for optimized distribution network;

[0016] The voltage at each node v i The deviation from the rated voltage v0 is used as the fitness function fit( ), and the specific expression is as follows:

[0017]

[0018] The constraints that need to be met during the optimization process are as follows:

[0019] Voltage upper and lower limit constraints:

[0020] v min ≤v i ≤v max ;

[0021] Power flow equation constraints:

[0022]

[0023] Among them, P gi ,Q gi are the active power output of the generator and the reactive power output of the generator respectively; P li ,Q li is the active power demand and reactive power demand of the jth node load; G ij , B ij are the conductance and susceptance parameters of the devices connected to the i-node and the j-node respectively; v i , v j are the voltages of node i and node j respectively; θ ij is the node voltage phase angle; v min 、v max are the lower and upper limits of node voltage respectively;

[0024] S5. By periodically monitoring the node voltage deviation, when the deviation is greater than a given threshold, the distribution network reconstruction is started.

[0025] Preferably, step S2 specifically includes the following sub-steps:

[0026] S21. Construct the probability density functions of distributed photovoltaic and load as G j ~P G (g j ) and C i ~P C (c i ), where i is the number of the i-th distributed photovoltaic and j is the load connected to the j-th busbar;

[0027] S22. Construct a calculation model for distributed photovoltaic output and obtain real-time distributed photovoltaic output. The calculation model for the active power injected into the grid by distributed photovoltaic is as follows:

[0028]

[0029] Among them, P DC-loss is the proportion of energy loss in the photovoltaic collection system; η inv Represents the efficiency of the power electronic converter; η t is the efficiency coefficient of the step-up transformer; P out-DC is the theoretical output of distributed photovoltaic power station, P out-AC Inject active power into the grid for distributed photovoltaics; P in-DC input power to the power electronic converter;

[0030] S23. Obtain real-time load demand through normal distribution.

[0031] Preferably, the method for constructing the optimization function in step S4 is as follows:

[0032] S41. Construct the initial population X of the disconnected lines in the distribution network p :

[0033] X p =[X p1 ,X p2 ,...,X pn ]

[0034] Among them, X p is the initial population of disconnected lines in the distribution network, which is obtained by representing the vector X of n disconnected lines, X = [X1, X2, ..., X n ], p∈[1,Np], Np is the population size;

[0035] S42, population mutation to obtain mutant population:

[0036] Disconnect the initial population X of lines from the distribution network p Randomly select two vectors X from a , X b , the difference between the two vectors is amplified and then compared with the individual X with the best fitness in the population * The sum is used to obtain the new mutant population. The calculation formula is as follows:

[0037] X′ p =X * +F(X a -X b )=[x′ p,1 ,x′ p,2 ,…,x′ p,n ]

[0038] Where F is the amplification factor, F∈[0,2];

[0039] S43, Genetic recombination: disconnect the line in the distribution network and the initial population X p With the mutant population X′ p Exchange information between them to obtain the disconnected line sub-population T of the distribution network p , T p =[T p1 ,T p2 ,...,T pn ], the formula for gene recombination is as follows:

[0040] t pi ={x′ pi / x pi ,if rand[0,1]<C r}

[0041] Among them, C r is a constant, representing the gene recombination rate, and i represents the i-th population;

[0042] S44, using the distribution network to disconnect the line initial population X p The line subpopulation T disconnected from the distribution network p Build an optimization function.

[0043] Preferably, the system parameters in step S1 include rated voltage, voltage limit, admittance matrix Y and initial grid topology X0, micro-phasor measurement unit and the location of distributed photovoltaic.

[0044] Preferably, in step S41, n=M-(m-1), m is the number of distribution network busbars, M is the number of interconnection lines, and m-1 is the number of closed-state lines.

[0045] Preferably, the representation of the i-th population in step S41 is:

[0046]

[0047] Among them, x max 、x min They are respectively the maximum and minimum line numbers in the distribution network.

[0048] Preferably, in step S4, the node voltage lower limit is set to 93% of the rated voltage, and the node voltage upper limit is set to 107% of the rated voltage.

[0049] Preferably, the normal distribution calculation formula in S23 is:

[0050]

[0051] Where f(x) is the probability density function, μ and σ are the mean and standard deviation respectively.

[0052] On the other hand, the present invention provides an optimization system for the above-mentioned distribution network optimization method considering the time-varying distributed power generation output and load demand, which comprises:

[0053] Initialization unit, used for initializing system parameters;

[0054] Collection unit, used to collect distributed photovoltaic output and load demand;

[0055] The missing point load acquisition unit is used to build a load state estimation model, obtain the measured missing point load based on the load state estimation model, and obtain the voltage and current of all nodes

[0056] An optimization unit, used for optimizing the distribution network reconstruction by using an optimization function;

[0057] The reconstruction unit is used to periodically monitor the node voltage deviation and start the distribution network reconstruction when the deviation is greater than a given threshold.

[0058] Finally, the present invention provides a distribution network optimization device that takes into account the time-varying distributed power generation output and load demand, which includes the above-mentioned optimization system.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] (1) The present invention proposes a distribution network optimization method that considers the time-varying distribution power output and load demand. The method is based on a differential evolution algorithm and can simultaneously consider the random changes of load and power output, and reconstruct and optimize the distribution network, thereby improving the voltage level and power loss of the distribution network. At the same time, the introduction of the state estimation method makes the method also applicable to the reconstruction problem of distribution networks in remote areas caused by insufficient coverage of measurement meters.

[0061] (2) After the network is optimized and reconstructed by the method of the present invention, the active power loss of the power grid can be greatly reduced. At the same time, the minimum node voltage and the average node voltage of the network after reconstruction are also significantly improved compared with those before reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0063] Figure 2 It is a schematic diagram of the system structure of the present invention;

[0064] Figure 3 The figure is a schematic diagram of a method flow of a specific embodiment of the present invention. DETAILED DESCRIPTION

[0065] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0066] Specifically, in a first aspect, the present invention provides a distribution network optimization method taking into account the time-varying distributed power generation output and load demand, such as Figure 1 As shown, it includes the following steps:

[0067] S1. Initialization of system parameters, where the system parameters include rated voltage, voltage limit, admittance matrix Y, initial grid topology X0, micro-phasor measurement unit and the location of distributed photovoltaics.

[0068] S2, collecting distributed photovoltaic output and load demand, specifically including the following sub-steps:

[0069] S21. Construct the probability density functions of distributed photovoltaic and load as G j ~P G (g j ) and Ci ~P C (c i ), where i is the number of the i-th distributed photovoltaic and j is the load connected to the j-th bus.

[0070] S22. Construct a calculation model for distributed photovoltaic output and obtain real-time distributed photovoltaic output. The calculation model for the active power injected into the grid by distributed photovoltaic is as follows:

[0071] P out-AC =(1-P DC-loss ) out-DC η inv η t ,

[0072] Among them, P DC-loss is the proportion of energy loss in the photovoltaic collection system; η inv Represents the efficiency of the power electronic converter; η t is the efficiency coefficient of the step-up transformer; P out-DC is the theoretical output of distributed photovoltaic power station, P out-AC Inject active power into the grid for distributed photovoltaics; P in-DC Input power to the power electronic converter.

[0073] S23. Obtain real-time load demand through normal distribution. The normal distribution calculation formula is:

[0074]

[0075] Where f(x) is the probability density function, μ and σ are the mean and standard deviation respectively.

[0076] S3. Construct a load state estimation model. Based on the load state estimation model, obtain the load of the missing point and obtain the voltage and current of all nodes. When the meter is missing, this method can be used to obtain the voltage and current of the missing point, so as to obtain the voltage and current of all nodes. This is very important for the problem of low meter coverage in some remote areas. The specific calculation formula is as follows:

[0077]

[0078] I u =V u Y u

[0079] Among them, I u To measure the missing node branch current value, V u To measure the missing node voltage estimate, V k To measure the normal node voltage value, I k To measure the normal node branch current value, Yk , Y ku , Y u are the elements in the system admittance matrix, where:

[0080] S4. Use the optimization function to optimize the distribution network reconstruction. The optimization function is as follows:

[0081] X p :={T p / X p ,if fit(T p )<fit(X p )}

[0082] Among them, fit( ) is the fitness function, X p is the initial population of disconnected lines in the distribution network, T p is the subpopulation after the gene combination of the disconnected lines in the distribution network; X p : The disconnected line population of the optimized distribution network.

[0083] The voltage at each node v i The deviation from the rated voltage v0 is used as the fitness function fit( ), and the specific expression is as follows:

[0084]

[0085] The constraints that need to be met during the optimization process are as follows:

[0086] Voltage upper and lower limit constraints:

[0087] v min ≤v i ≤v max .

[0088] Power flow equation constraints:

[0089]

[0090] Among them, P gi ,Q gi are the active power output of the generator and the reactive power output of the generator respectively; P li ,Q li is the active power demand and reactive power demand of the jth node load; G ij , B ij are the conductance and susceptance parameters of the devices connected to the i-node and the j-node respectively; v i , v j are the voltages of node i and node j respectively; θ ij is the node voltage phase angle; v min 、v maxIn this embodiment, the node voltage lower limit is set to 93% of the rated voltage, and the node voltage upper limit is set to 107% of the rated voltage.

[0091] Preferably, the method for constructing the optimization function in step S4 is as follows:

[0092] S41. Construct the initial population X of the disconnected lines in the distribution network p :

[0093] X p =[X p1 ,X p2 ,...,X pn ]

[0094] Among them, X p is the initial population of disconnected lines in the distribution network, which is obtained by representing the vector X of n disconnected lines, X = [X1, X2, ..., X n ], p∈[1,Np], Np is the population size.

[0095] Preferably, in step S41, n=M-(m-1), m is the number of distribution network buses, M is the number of interconnection lines, and m-1 is the number of closed-state lines. The representation of the i-th population is:

[0096]

[0097] Among them, x max 、x min They are respectively the maximum and minimum line numbers in the distribution network.

[0098] S42, population mutation to obtain mutant population:

[0099] Disconnect the initial population X of lines from the distribution network p Randomly select two vectors X from a , X b , the difference between the two vectors is amplified and then compared with the individual X with the best fitness in the population * The sum is used to obtain the new mutant population. The calculation formula is as follows:

[0100] X′ p =X * +F(X a -X b )=[x′ p,1 ,x′ p,2 ,…,x′ p,n ]

[0101] Where F is the amplification factor, F∈[0,2].

[0102] S43, Genetic recombination: disconnect the line in the distribution network and the initial population Xp With the mutant population X′ p Exchange information between them to obtain the disconnected line sub-population T of the distribution network p , T p =[T p1 ,T p2 ,...,T pn ], the formula for gene recombination is as follows:

[0103] t pi ={x′ pi / x pi ,if rand[0,1]<C r}

[0104] Among them, C r is a constant representing the gene recombination rate, and i represents the i-th population.

[0105] S44, using the distribution network to disconnect the line initial population X p The line subpopulation T disconnected from the distribution network p Build an optimization function.

[0106] S5. By periodically monitoring the node voltage deviation, when the deviation is greater than a given threshold, the distribution network reconstruction is initiated. In a specific embodiment, the distribution network reconstruction is generally achieved by opening and closing switches to optimize the interconnection relationship.

[0107] On the other hand, the present invention provides an optimization system for the above-mentioned distribution network optimization method considering the time-varying distributed power generation output and load demand, such as Figure 2 As shown, it includes:

[0108] Initialization unit 1 is used for initializing system parameters.

[0109] The collection unit 2 is used to collect distributed photovoltaic output and load demand.

[0110] The missing point load acquisition unit 3 is used to construct a load state estimation model, obtain the measured missing point load based on the load state estimation model, and obtain the voltage and current of all nodes.

[0111] The optimization unit 4 is used to optimize the distribution network reconstruction by using the optimization function.

[0112] The reconstruction unit 5 is used to periodically monitor the node voltage deviation and start the distribution network reconstruction when the deviation is greater than a given threshold.

[0113] Finally, the present invention provides a distribution network optimization device that takes into account the time-varying distributed power generation output and load demand, which includes the above-mentioned optimization system. Specific embodiments

[0115] The flow chart of the distribution network reconstruction strategy in this embodiment is as follows Figure 3 As shown, the specific steps include:

[0116] S1. System parameter initialization. System parameter initialization includes setting rated voltage, voltage limit, admittance matrix Y, initial grid topology X0, micro-phasor measurement unit μPMU and the location of distributed photovoltaics. It should be noted that completely random data cannot be used as the initial value of the algorithm during the initialization process. The current topology needs to be included in the random population so that when the algorithm cannot find a better solution, the grid remains in its original state.

[0117] S2. Sampling of distributed photovoltaic output and load demand. Assume that Ci and Gj are random variables representing load demand and power generation output, respectively, and their probability density functions are Ci~PC(ci), Gj~PG(gj), respectively, where i and j are the load connected to the i-th bus and the j-th distributed photovoltaic number. The probability density functions of load demand and power generation output are independent of each other and satisfy normal distribution.

[0118] The theoretical output of a distributed photovoltaic power station is:

[0119]

[0120] Among them, I represents solar irradiance (W / m2), Ibase represents benchmark irradiance, usually 1000W / m2, and Pmax represents distributed photovoltaic installation capacity.

[0121] The efficiency of a solar power generation unit is generally inversely proportional to the temperature Tcell. Based on this, formula (1) can be further improved to formula (2). Wherein, the reference temperature Tref = 25°C, and the temperature coefficient γ is -0.5% / °C.

[0122]

[0123] After considering DC loss and transmission efficiency, the active power injected into the grid can be expressed as follows:

[0124] P out-AC =(1-P DC-loss ) out-DC η inv η t ,

[0125] Among them, P DC-loss is the proportion of energy loss in the photovoltaic collection system; η inv Represents the efficiency of the power electronic converter; η t is the efficiency coefficient of the step-up transformer.

[0126] S3. Estimation of load state at measurement missing points. To address the problem of incomplete measurement coverage of the distribution network, the present invention proposes to adopt a state estimation algorithm to use the voltage Vk and current Ik data measured by the existing μPMU device to estimate the voltage data of other nodes, thereby forming the voltage vector data of all nodes in the entire network. The specific algorithm is as follows.

[0127] According to Kirchhoff's law, we have:

[0128] YV=I (4)

[0129] Where Y is the system admittance matrix.

[0130] The above formula can be further expressed as follows:

[0131]

[0132] The subscript k represents a known measurement value, and the subscript u represents an unknown measurement value.

[0133] From formula (5), we can get:

[0134]

[0135] Where I is the identity matrix.

[0136] The state estimation value of the node voltage can be obtained from formula (6) as follows:

[0137]

[0138] After obtaining the estimated value of each node voltage, the current value of each branch can be calculated by I=YV. After obtaining the system voltage and current data, the load data of each node can be calculated by S=VI*.

[0139] S4. Use the optimization function to optimize the distribution network reconstruction. The optimization function is as follows:

[0140] X p :={T p / X p ,if fit(T p )<fit(X p )}

[0141] Among them, fit( ) is the fitness function, X p is the initial population of disconnected lines in the distribution network, T p is the subpopulation after the gene combination of the disconnected lines in the distribution network; X p : The disconnected line population of the optimized distribution network.

[0142] The voltage at each node v iThe deviation from the rated voltage v0 is used as the fitness function fit( ), and the specific expression is as follows:

[0143]

[0144] The constraints that need to be met during the optimization process are as follows:

[0145] Voltage upper and lower limit constraints:

[0146] v min ≤v i ≤v max .

[0147] Power flow equation constraints:

[0148]

[0149] Among them, P gi ,Q gi are the active power output of the generator and the reactive power output of the generator respectively; P li ,Q li is the active power demand and reactive power demand of the jth node load; G ij , B ij are the conductance and susceptance parameters of the devices connected to the i-node and the j-node respectively; v i , v j are the voltages of node i and node j respectively; θ ij is the node voltage phase angle; v min 、v max In this embodiment, the node voltage lower limit is set to 93% of the rated voltage, and the node voltage upper limit is set to 107% of the rated voltage.

[0150] Preferably, the method for constructing the optimization function in step S4 is as follows:

[0151] S41. Construct the initial population X of the disconnected lines in the distribution network p :

[0152] X p =[X p1 ,X p2 ,...,X pn ]

[0153] Among them, X p is the initial population of disconnected lines in the distribution network, which is obtained by representing the vector X of n disconnected lines, X = [X1, X2, ..., X n ], p∈[1,Np], Np is the population size.

[0154] Preferably, in step S41, n=M-(m-1), m is the number of distribution network buses, M is the number of interconnection lines, and m-1 is the number of closed-state lines. The representation of the i-th population is:

[0155]

[0156] Among them, x max 、x min They are respectively the maximum and minimum line numbers in the distribution network.

[0157] S42, population mutation to obtain mutant population:

[0158] Disconnect the initial population X of lines from the distribution network p Randomly select two vectors X from a , X b , the difference between the two vectors is amplified and then compared with the individual X with the best fitness in the population * The sum is used to obtain the new mutant population. The calculation formula is as follows:

[0159] X′ p =X * +F(X a -X b )=[x′ p,1 ,x′ p,2 ,…,x′ p,n ]

[0160] Where F is the amplification factor, F∈[0,2]. The individual X with the best fitness * It is the individual with the smallest fitness function value.

[0161] S43, Genetic recombination: disconnect the line in the distribution network and the initial population X p With the mutant population X′ p Exchange information between them to obtain the disconnected line sub-population T of the distribution network p , T p =[T p1 ,T p2 ,...,T pn ], the formula for gene recombination is as follows:

[0162] t pi ={x′ pi / x pi ,if rand[0,1]<C r}

[0163] Among them, C r is a constant representing the gene recombination rate, and i represents the i-th population.

[0164] S44, using the distribution network to disconnect the line initial population X pThe line subpopulation T disconnected from the distribution network p Construct the optimization function. The construction process is completed by gene screening, which compares the original vector X p With the test vector T p The fitness between them is used to evaluate the quality of the results. Among them, only the vector with the best fitness, that is, the vector with the lowest fitness function solution value, is selected for the next round of iterative optimization.

[0165] S5. By periodically monitoring the node voltage deviation, when the deviation is greater than a given threshold, the distribution network reconstruction is initiated, thereby forming an event triggering mechanism for network reconstruction, thereby optimizing the entire distribution network. In specific applications, the given threshold is generally given based on a database or empirical value.

[0166] In this embodiment, the actual distribution network system of a certain province is used as an example system to verify the effectiveness of the distribution network optimization method proposed by this method. The power grid system includes 33 buses and 37 distribution lines. Among them, the population number of the differential evolution algorithm is set to 300, the maximum number of iterations is 200 times, the gene recombination rate is 0.7, and the amplification factor is 0.5.

[0167] Optimization results without considering distributed photovoltaic access scenarios:

[0168] When the distributed photovoltaic power generation is not considered, the effectiveness of the proposed method is verified by comparing the differences in power loss, minimum voltage, average voltage and other indicators between the actual distribution network and the reconstructed distribution network. The reconstruction results are shown in Table 1.

[0169] Table 1 Reconstruction results without considering distributed photovoltaic power generation scenario

[0170]

[0171] As shown in Table 1, after the network reconstruction, the line number is optimized from [33, 34, 35, 36, 37] to [11, 14, 27, 33, 36], and the active power loss is reduced from 202.49kW to 153.23kW, a decrease of 24%. At the same time, the minimum node voltage and average node voltage of the network after reconstruction are also significantly improved compared with those before reconstruction.

[0172] Furthermore, considering the network reconstruction results under different load demand conditions, three scenarios of 50%, 75% and 115% are analyzed respectively, and the results are shown in Table 2.

[0173] Table 2 Reconstruction results under different load demand scenarios

[0174]

[0175]

[0176] As can be seen from Table 2, under different load demand levels, the proposed method can not only improve the minimum voltage level, but also significantly improve the average voltage level of the network. At the same time, the active power loss of the network also decreased, with an average decrease of 26%. By comparing the reconstruction results under different scenarios, it can be seen that the key lines that need to be disconnected are 14, 17 and 28.

[0177] Consider the optimization results under the distributed photovoltaic access scenario:

[0178] When considering the access of distributed photovoltaic power plants, it is assumed that five distributed photovoltaic power plants are connected to nodes 5, 8, 10, 25, and 33, with capacities of 500kW, 250kW, 250kW, 350kW, and 125kW, respectively. Considering the three scenarios of 100%, 85%, and 65%, the reconstruction results are shown in Table 3.

[0179] Table 3 Reconstruction results of distributed photovoltaic under different output scenarios

[0180]

[0181] As can be seen from Table 3, as the output of distributed photovoltaics increases, the average node voltage and the minimum node voltage also gradually increase, proving the voltage support effect of distributed photovoltaics on the power grid. At the same time, after the network reconstruction, the average power loss of the entire network decreased by 26%. The optimization results also prove the feasibility of obtaining the optimal operation point of the power grid by optimizing the output of distributed photovoltaics and combining the network reconstruction method.

[0182] Optimization results considering the scenario of simultaneous changes in load and distributed photovoltaics:

[0183] In the actual distribution network, the distributed photovoltaic output and load demand are in real-time dynamic change process, which is taken into consideration in this section.

[0184] The distributed photovoltaic power generation output is described in Section 1.2 and follows the typical solar irradiation characteristics. The mean square error of the load demand probability density function is taken as 0.1pu, that is, in 95% of the time, the load demand fluctuates within the range of ±0.2pu of the rated value. The reconstruction results are shown in Table 4.

[0185] Table 4 Reconstruction results considering the scenario of simultaneous changes in load and distributed photovoltaics

[0186]

[0187]

[0188] As can be seen from Table 4, after the network reconstruction, in each time period, the node voltage amplitude increased to a certain extent, and the active power loss also decreased. The higher the load demand, the greater the improvement in voltage and active power loss. Before and after the network reconstruction, the total active power loss decreased from 7456.07kW to 5591.21kW, a decrease of 25%.

[0189] The embodiments described above are only descriptions of the preferred implementation modes of the present invention, and are not intended to limit the scope of the present invention. Without departing from the design spirit of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the protection scope determined by the claims of the present invention.

Claims

1. A distribution network optimization method considering the time-varying distributed power generation output and load demand, characterized by: It includes the following steps: S1, system parameter initialization; S2, collect distributed photovoltaic output and load demand; S3. Construct a load state estimation model. Based on the load state estimation model, obtain the load of the measurement missing point and obtain the voltage and current of all nodes. The load state estimation model is as follows: I u =V u Y u Among them, I u To measure the missing node branch current value, V u To measure the missing node voltage estimate, V k To measure the normal node voltage value, I k To measure the normal node branch current value, Y k , Y ku , Y u are the elements in the system admittance matrix respectively, where: S4. Use the optimization function to optimize the distribution network reconstruction. The optimization function is as follows: X p :={T p / X p ,iffit(T p )<fit(X p )} Among them, fit() is the fitness function, X p is the initial population of disconnected lines in the distribution network, T p is the subpopulation after the gene combination of the disconnected lines in the distribution network; X p : Disconnect line populations for optimized distribution network; The voltage at each node v i The deviation from the rated voltage v0 is used as the fitness function fit(), and the specific expression is as follows: The constraints that need to be met during the optimization process are as follows: Voltage upper and lower limit constraints: in min ≤in i ≤in max ; Power flow equation constraints: Among them, P gi ,Q gi are the active power output of the generator and the reactive power output of the generator respectively; P li ,Q li is the active power demand and reactive power demand of the j-th node load; G ij , B ij are the conductance and susceptance parameters of the devices connected to the i-node and the j-node respectively; v i , v j are the voltages of node i and node j respectively; θ ij is the node voltage phase angle; v min 、v max are the lower and upper limits of node voltage respectively; S5. By periodically monitoring the node voltage deviation, when the deviation is greater than a given threshold, the distribution network reconstruction is started.

2. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: Step S2 specifically includes the following sub-steps: S21. Construct the probability density functions of distributed photovoltaic and load as G j ~P G (g j ) and C i ~P C (c i ), where i is the number of the i-th distributed photovoltaic and j is the load connected to the j-th busbar; S22. Construct a calculation model for distributed photovoltaic output and obtain real-time distributed photovoltaic output. The calculation model for the active power injected into the grid by distributed photovoltaic is as follows: P out-AC =(1-P DC-loss )P out-DC or inv or t , Among them, P DC-loss is the proportion of energy loss in the photovoltaic collection system; η inv Represents the efficiency of the power electronic converter; η t is the efficiency coefficient of the step-up transformer; P out-DC is the theoretical output of distributed photovoltaic power station, P out-AC Inject active power into the grid for distributed photovoltaics; P in-DC input power to the power electronic converter; S23. Obtain real-time load demand through normal distribution.

3. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: The construction method of the optimization function in step S4 is as follows: S41. Construct the initial population X of the disconnected lines in the distribution network p : X p =[X p1 ,X p2 ,...,X pn ] Among them, X p is the initial population of disconnected lines in the distribution network, which is obtained by representing the vector X of n disconnected lines, X = [X1, X2, ..., X n ], p∈[1,Np], Np is the population size; S42, population mutation to obtain mutant population: Disconnect the initial population X of lines from the distribution network p Randomly select two vectors X from a , X b , the difference between the two vectors is amplified and then compared with the individual X with the best fitness in the population * The sum is used to obtain the new mutant population. The calculation formula is as follows: X′ p =X * +F(X a -X b )=[x′ p,1 ,x′ p,2 ,…,x′ p,n ] Where F is the amplification factor, F∈[0,2]; S43, Genetic recombination: disconnect the line in the distribution network and the initial population X p With the mutant population X′ p Exchange information between them to obtain the disconnected line sub-population T of the distribution network p , T p =[T p1 ,T p2 ,...,T pn ], the formula for gene recombination is as follows: t pi ={x′ pi / x pi ,ifrand[0,1]<C r } Among them, C r is a constant, representing the gene recombination rate, and i represents the i-th population; S44, using the distribution network to disconnect the line initial population X p The line subpopulation T is disconnected from the distribution network p Build an optimization function.

4. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: The system parameters in step S1 include rated voltage, voltage limit, admittance matrix Y and initial grid topology X0, micro-phasor measurement unit and the location of distributed photovoltaic.

5. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: In step S41, n=M-(m-1), m is the number of distribution network buses, M is the number of interconnection lines, and m-1 is the number of closed-state lines.

6. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: The representation of the i-th population in step S41 is: x p,i =[x min +rand[0,1](x max -x min )]=[Mrand[0,1]] Among them, x max 、x min They are respectively the maximum and minimum line numbers in the distribution network.

7. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: In step S4, the node voltage lower limit is set to 93% of the rated voltage, and the node voltage upper limit is set to 107% of the rated voltage.

8. The method for optimizing a distribution network considering time-varying distributed power generation output and load demand according to claim 1, characterized in that: The normal distribution calculation formula in S23 is: Where f(x) is the probability density function, μ and σ are the mean and standard deviation respectively.

9. An optimization system for the distribution network optimization method considering the time-varying distributed power generation output and load demand as claimed in claim 1, characterized in that: It includes: Initialization unit, used for initializing system parameters; Collection unit, used to collect distributed photovoltaic output and load demand; The missing point load acquisition unit is used to build a load state estimation model, obtain the measured missing point load based on the load state estimation model, and obtain the voltage and current of all nodes An optimization unit, used for optimizing the distribution network reconstruction by using an optimization function; The reconstruction unit is used to periodically monitor the node voltage deviation and start the distribution network reconstruction when the deviation is greater than a given threshold.

10. A distribution network optimization device that takes into account the time-varying distributed power generation output and load demand, characterized in that: It comprises the system as claimed in claim 9.

Images