Global Optimization Method for Power Quality of Microgrid

Abstract

A method for global optimization of the power quality of a microgrid, the method comprising: establishing a node admittance matrix of the microgrid based on the equivalent virtual impedance of the inverters at each node of the microgrid and the local loads at each node (S1); constructing an objective function of the microgrid based on the weight values assigned to each node in the microgrid and the voltage distortion rate of each node (S2); constructing a constraint condition for harmonic modulation of each node of the microgrid based on the node admittance matrix and the remaining capacity of the inverters at each node (S3); in response to the constraint condition being satisfied, analyzing the optimal solution of the objective function and adjusting the equivalent virtual impedance at each node based on the optimal solution (S4).

Description

Technical Field

[0001] This application claims the priority of a Chinese patent application with the application number 202310878580.3, which was filed with the Chinese Patent Office on July 18, 2023, and all the contents of that application are incorporated herein by reference.

[0002] This application relates to the technical field of power supply, for example, to a method for global optimization of power quality in a microgrid.

Background Art

[0003] A microgrid is a small power grid that includes distributed energy, loads, etc., has two operating modes of island and grid-connected, and can achieve flexible switching. With the requirement of local consumption of distributed energy, the construction of microgrids has also entered a stage of rapid development. With the increase in the number and power of nonlinear loads in microgrids, the problem of voltage harmonic distortion rate in microgrids has become prominent.

[0004] Currently, for harmonic processing in microgrids, most measures mainly involve using specialized harmonic processing equipment such as active filters similar to those in conventional distribution grids. However, currently, since microgrids are mainly used to solve the power demand in remote areas, adding additional specialized voltage quality processing equipment is restricted by factors such as economic costs.

Summary of the Invention

[0005] Embodiments of this application provide a method for global optimization of power quality in a microgrid. In the current microgrid, the problem of harmonics in the microgrid is solved by additionally installing specialized voltage quality processing equipment, but that method is restricted by factors such as economic costs and is not suitable for the power demand in remote areas, which is the problem to be solved.

[0006] On one side, this application Establishing the node admittance matrix of the microgrid based on the equivalent virtual impedance of the inverters of each node in the microgrid and the local loads of each node; Constructing the objective function of the microgrid based on the weight values assigned to each node in the microgrid and the voltage distortion rate of each node; Constructing the constraint conditions for harmonic modulation of each node in the microgrid based on the node admittance matrix and the remaining capacity of the inverters of each node; In response to the satisfaction of the constraint conditions, analyzing the objective function corresponding to each particle in the particle swarm, determining the optimal solution of the objective function, and adjusting the equivalent virtual impedance of the inverters of each node based on the optimal solution, wherein one set of values of the real part and the imaginary part of the equivalent virtual impedance at all nodes in the microgrid forms one of the particles; Providing a method for global optimization of the power quality of a microgrid.

[0007] In some embodiments, constructing the objective function of the microgrid based on the weight values assigned to each node in the microgrid and the voltage distortion rate of each node as described above includes: Determining the voltage distortion rate of each node based on the effective value of the fundamental voltage and the effective value of the harmonic voltage of each node; Multiplying and then adding the voltage distortion rate of each node and the weight value of each node in correspondence to obtain the objective function of the microgrid.

[0008] In some embodiments, the objective function of the microgrid is: JPEG0007714817000002.jpg1338 wherein: N represents the number of nodes in the microgrid; JPEG0007714817000003.jpg66 is the weight value corresponding to the j-th node; JPEG0007714817000004.jpg612 is the voltage distortion rate corresponding to the j-th node, In some embodiments, the voltage distortion rate is, JPEG0007714817000005.jpg2660 where, Among them, JPEG0007714817000006.jpg68 is the effective value of the fundamental voltage at the j-th node, JPEG0007714817000007.jpg76 is the effective value of the h-th harmonic voltage at the j-th node, JPEG0007714817000008.jpg728, JPEG0007714817000009.jpg66 is the node admittance matrix for the h-th harmonic, JPEG0007714817000010.jpg74 is the h-th harmonic current at the j-th node.

[0009] In some embodiments, constructing the constraint conditions for harmonic modulation of each node of the microgrid based on the above-mentioned node admittance matrix and the residual capacitance of the inverter at each node includes: determining the residual capacitance of the inverter based on the rated capacitance of the inverter, the active power output by the inverter, and the reactive power output by the inverter; determining the compensation capacitance of the inverter based on the fundamental voltage of the inverter and the harmonic currents of each order flowing through the inverter, wherein the constraint conditions include that the compensation capacitance is smaller than the residual capacitance.

[0010] In some embodiments, the residual capacitance is, JPEG0007714817000011.jpg1049 where, Among them, JPEG0007714817000012.jpg66 is the residual capacitance of the inverter, JPEG0007714817000013.jpg76 is the rated capacity of the inverter, P is the active power output by the inverter, Q is the reactive power output by the inverter, The compensation capacitance is JPEG0007714817000014.jpg1552 and Among them, JPEG0007714817000015.jpg66 is the compensation capacitance of the inverter, JPEG0007714817000016.jpg613 is the fundamental voltage of the inverter, JPEG0007714817000017.jpg613 is the harmonic current flowing through the inverter.

[0011] In some embodiments, in response to the above-described satisfaction of the constraint conditions, analyzing the objective function corresponding to each particle in the particle set, determining the optimal solution of the objective function, and adjusting the equivalent virtual impedance of the inverter of each node based on the optimal solution includes analyzing the objective function corresponding to each of the particles in the particle set, comparing the objective functions corresponding to all of the particles in the particle set, determining the particle corresponding to the minimum objective function, and designating the particle as the global optimal particle; and inputting an initial particle, an initial iteration speed, and an initial optimal particle randomly generated into a particle update model for updating the next particle with the previous particle and the iteration speed of the previous particle, and updating the first particle in the particle set; comparing the objective function of the updated first particle with the objective function of the initial optimal particle; in response to the objective function of the updated first particle being smaller than the objective function of the initial optimal particle, designating the updated first particle as the historical optimal particle instead of the initial optimal particle; Input the updated particles, the history-optimal particles, and the current iteration speed into the particle update model to obtain the next iteration speed and the next updated particle, and sequentially and cyclically update all the particles in the particle set. After each update, compare the objective function corresponding to the updated particles with the history-optimal particles. In response to the objective function corresponding to the updated particles being smaller than the objective function corresponding to the history-optimal particles, replace the history-optimal particles with the updated particles. Compare the objective function corresponding to the updated particles with the objective function corresponding to the global-optimal particles. In response to the objective function corresponding to the updated particles being smaller than the objective function corresponding to the global-optimal particles, replace the global-optimal particles with the updated particles. After all the particles in the particle set are updated, one iteration update of the particle set is completed. Update the particle set a predetermined number of times to obtain the final global-optimal particles. Based on the global-optimal particles, adjust the equivalent virtual impedance at each node in the microgrid.

[0012] In some embodiments, the above-mentioned step of comparing the objective function corresponding to the updated particles with the history-optimal particles after each update includes: After each update, determine whether the value of the particles is within a predetermined range. In response to the value of the particles exceeding the predetermined range, end the cyclic update. In response to the value of the particles being within the predetermined range, compare the objective function corresponding to the updated particles with the history-optimal particles.

[0013] In some embodiments, after the above-mentioned step of comparing the objective function corresponding to the updated particles with the history-optimal particles after each update, the method for global optimization of the power quality of the microgrid includes: In response to the objective function corresponding to the updated particle being greater than or equal to the objective function corresponding to the historical optimal particle, mark the update of this time as a failed update, and count the number of consecutive failed updates; In response to the number of consecutive failed updates being less than the failure count threshold, continue the cyclic update; In response to the number of consecutive failed updates being greater than or equal to the failure count threshold, convert the current particle to the learning function mode, obtain the particle after learning, reset the number of consecutive failed updates, replace the current particle with the particle after learning, and re-enter the cyclic update; In response to the objective function corresponding to the updated particle being less than the objective function corresponding to the historical optimal particle, further include resetting the number of consecutive failed updates.

[0014] In some embodiments, in response to the number of consecutive failed updates being greater than or equal to the failure count threshold, converting the current particle to the learning function mode, obtaining the particle after learning, resetting the number of consecutive failed updates, replacing the current particle with the particle after learning, and re-entering the cyclic update, includes: In response to the number of consecutive failed updates being greater than or equal to the failure count threshold, generate a random number, and determine whether the random number is less than a learning probability factor calculated based on the position number of the current particle in the particle set and the total number of particles in the particle set; In response to the random number being less than the learning probability factor, randomly select n non-updated particles behind the current particle in the particle set; Compare the objective functions corresponding to the selected n particles, select the particle corresponding to the minimum objective function among them, mark the particle as the learning optimal particle, and replace the current particle with the learning optimal particle; In response to the random number being greater than or equal to the learning probability factor, replacing each piece of data in the current particle with the data that is optimal for the history of the individual of that data; resetting the continuous update failure count, and then reintroducing the updated current particle into cyclic updating.

[0015] In some embodiments, after comparing the objective function corresponding to the updated particle with the objective function corresponding to the global optimal particle as described above, the global optimization method for the power quality of the microgrid is as follows: further including maintaining the global optimal particle as it is in response to the objective function corresponding to the updated particle being greater than or equal to the objective function corresponding to the global optimal particle.

[0016] On the other hand, the present application provides: a node admittance matrix construction module configured to establish a node admittance matrix of the microgrid based on the equivalent virtual impedance of the inverters of each node of the microgrid and the local load of each node; an objective function establishment module configured to construct an objective function of the microgrid based on the weight values assigned to each node in the microgrid and the voltage distortion rate of each node; a constraint condition construction module configured to construct constraint conditions for harmonic modulation of each node of the microgrid based on the node admittance matrix and the remaining capacity of the inverters of each node; an optimal solution search module configured to analyze the objective function corresponding to each particle in the particle set in response to the constraint conditions being satisfied, determine the optimal solution of the objective function, and adjust the equivalent virtual impedance of the inverters of each node based on the optimal solution, wherein one set of values of the real part and the imaginary part of the equivalent virtual impedance at all nodes in the microgrid constitutes one of the particles. A system for globally optimizing the power quality of a microgrid is provided.

Brief Description of the Drawings

[0017]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Modes for Carrying Out the Invention

[0018] Hereinafter, specific embodiments of the present application will be described with reference to the drawings.

[0019] It should be noted that the terms "first", "second", etc. in the specification, claims and above drawings of the present application do not necessarily need to be used to explain a specific order or sequence, but are for distinguishing similar objects. The data used in this way can be replaced when appropriate, and it should be understood that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0020] In the description of this specification, the description of reference terms such as "embodiment", "one embodiment", "one implementation form", etc. means that the features, structures, materials or characteristics described in connection with the embodiment or implementation form are included in at least one embodiment or implementation form of this application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or implementation form. Furthermore, the described features, structures, materials or characteristics can be combined in a suitable manner in any one or more embodiments or implementation forms.

[0021] Figure 1 shows a schematic diagram of the flow of a method for global optimization of the power quality of a microgrid according to an embodiment of this application. The method for global optimization of the power quality of the microgrid includes the following.

[0022] In S1, based on the equivalent virtual impedance of the inverters at each node of the microgrid and the local load at each node, establish the node admittance matrix of the microgrid. Exemplarily, the power grid is a group consisting of static elements such as transmission lines, power transformers, and shunt (series) capacitors. The node admittance matrix is one of the description forms for the power grid. Equivalent virtual impedance control is to introduce a control part of a virtual impedance into the current or voltage control circuit of the inverter to change the characteristics of the output voltage and current of the inverter, so as to represent the inverter as a power supply device with a physical impedance connected in series. By utilizing the existing inverters in the microgrid, an effective treatment of the harmonic problems in the microgrid can be realized, and the cost expenditure due to the additional installation of related power quality treatment equipment can be avoided.

[0023] In this embodiment, the relationship between the node admittance matrix Y, the harmonic voltage U, and the harmonic current I is YU = I.

[0024] The node admittance matrix at each harmonic is JPEG0007714817000018.jpg66, The harmonic current at the h - th harmonic of the j - th node in JPEG0007714817000019.jpg66 is JPEG0007714817000020.jpg86, and the harmonic voltage is JPEG0007714817000021.jpg77, and the relationship among the three is JPEG0007714817000022.jpg829. The micro - grid is restricted by harmonic power flow during the operation process. When establishing the equivalent circuit network of harmonics, for example, JPEG0007714817000023.jpg49103 the node admittance matrix at each harmonic as shown can be obtained, Among them, JPEG0007714817000024.jpg86 is the phasor representation of the harmonic voltage at the h - th harmonic of the j - th node, JPEG0007714817000025.jpg86 is the phasor representation of the harmonic current at the h - th harmonic of the j - th node, JPEG0007714817000026.jpg77 is the inherent impedance (i.e., local load) at the h - th harmonic of the j - th node when no virtual impedance is given, JPEG0007714817000027.jpg86 is the given equivalent virtual impedance.

[0025] In S2, based on the weight values assigned to each node in the micro - grid and the voltage distortion rate of each node, the objective function of the micro - grid is constructed, Exemplarily, for different nodes, there may be differences in the requirements for power quality. Therefore, by assigning different weight coefficients to different nodes, a differentiated processing effect of voltage harmonic distortion rate can be realized. When the processing capacity of the inverter is limited, the requirements for the power quality of important nodes are preferentially satisfied to improve the satisfaction of the users of the system with electricity use. To minimize the interference of harmonics, the objective function is constructed by multiplying and then adding the weight value of each node and the voltage distortion rate of the node in correspondence, and the voltage distortion rate of each node can be comprehensively considered in the objective function.

[0026] In S3, based on the node admittance matrix and the remaining capacity of the inverter at each node, the constraint conditions for harmonic modulation of each node of the microgrid are constructed. Exemplarily, in the process of the inverter performing harmonic compensation, its compensation capacity is restricted by the remaining capacity of the inverter, and the compensation capacity shall not exceed the remaining capacity, and the inverter shall not be overloaded. Also, the compensation capacity of the inverter needs to be calculated based on the harmonic current, and the harmonic current is related to the node admittance matrix.

[0027] In S4, when the constraint conditions are satisfied, the objective function corresponding to each particle in the particle set is analyzed, the optimal solution of the objective function is determined, and the equivalent virtual impedance of the inverter at each node is adjusted based on the optimal solution. Among them, a set of values of one type of the real part and the imaginary part of the equivalent virtual impedance at all nodes in the microgrid forms one of the particles.

[0028] In this embodiment, based on the node admittance matrix and the remaining capacitance of the inverter at each node, the constraint conditions for harmonic modulation of each node in the microgrid are constructed. With the equivalent virtual impedance as a variable in the node admittance matrix, weight values are assigned to each node in the microgrid. Considering the differences in requirements for power quality of different nodes, an objective function of the microgrid is constructed based on the weight values and the voltage distortion rate of each node. Aiming to minimize the value of the objective function, when the constraint conditions are satisfied, the optimal solution of the objective function is analyzed, and the equivalent virtual impedance of the inverter at each node is adjusted based on the optimal solution, thereby realizing the global optimization process of differential voltage harmonic distortion, without the need to additionally install specialized power quality processing equipment. By constructing the inverter and the equivalent virtual impedance of the inverter, effective harmonic processing can be realized.

[0029] Note that in some embodiments, if there is no inverter at a node, that is, if the inverter is not connected, there is no equivalent virtual impedance of the inverter at that node. Therefore, it is not necessary to adjust the equivalent virtual impedance of the inverter at that node as a variable.

[0030] In one embodiment of the present application, constructing the objective function of the microgrid based on the weight values assigned to each node in the microgrid and the voltage distortion rate of each node described above means that it includes determining the voltage distortion rate of each node based on the effective value of the fundamental voltage and the effective value of the harmonic voltage of each node. Among them, the voltage distortion rate (also called the voltage harmonic distortion rate) is an index to measure the degree of harmonic pollution in the power system. The voltage distortion rate represents the relative magnitude between the harmonic component and the fundamental wave component in the voltage waveform. Harmonic pollution causes hazards to the power system such as a decrease in equipment efficiency, overheating of equipment, interference with communication systems, an impact on the accuracy of watt-hour meters, and malfunction of protection devices.

[0031] Exemplarily, the voltage distortion rate is JPEG0007714817000028.jpg2660 and among them JPEG0007714817000029.jpg68 is the effective value of the fundamental voltage at the j-th node, JPEG0007714817000030.jpg76 is the effective value of the h-th harmonic voltage at the j-th node, h represents the order of the harmonic distortion, starting from the third harmonic distortion, calculating up to the ninth harmonic distortion, and calculating the voltage distortion rates corresponding to the third, ninth, and odd harmonic distortions between them.

[0032] Multiply the voltage distortion rate of each node by the corresponding weight value of each node and then sum them to obtain the objective function of the microgrid.

[0033] Exemplarily, the objective function of the microgrid is JPEG0007714817000031.jpg1338 and among them, N represents the number of nodes of the microgrid, JPEG0007714817000032.jpg56 is the weight value corresponding to the j-th node, and the sum of the weight coefficients of each node is 1, JPEG0007714817000033.jpg611 is the voltage distortion rate corresponding to the j-th node.

[0034] In one embodiment of the present application, constructing the constraint conditions for the harmonic modulation of each node of the microgrid based on the node admittance matrix and the remaining capacity of the inverter of each node described above is including determining the remaining capacity of the inverter based on the rated capacity of the inverter, the active power output by the inverter, and the reactive power output by the inverter, Exemplarily, the remaining capacity is JPEG0007714817000034.jpg1049 and among them JPEG0007714817000035.jpg76 is the remaining capacity of the inverter, JPEG0007714817000036.jpg66 is the rated capacity of the inverter, P is the active power output by the inverter, and Q is the reactive power output by the inverter.

[0035] Based on the fundamental voltage of the inverter and the harmonic currents of each order flowing through the inverter, determine the compensation capacitance of the inverter, Exemplarily, the compensation capacitance is, JPEG0007714817000037.jpg1552 and, among them, JPEG0007714817000038.jpg66 is the compensation capacitance of the inverter of the distributed power source, JPEG0007714817000039.jpg613 is the fundamental voltage of the inverter. In the optimization process, the fundamental voltage is approximately considered not to be affected, and the fundamental voltage of the capacitor connected in parallel to each inverter when no control is applied can be selected as the fundamental voltage in the optimization process. Therefore, the result calculated using this fundamental voltage can be approximately regarded as the compensation capacitance. JPEG0007714817000040.jpg713 is the harmonic current flowing through the inverter. The compensation variable can be calculated by substituting the fundamental voltage and harmonic current of the inverter at any node into the formula, where h = 3, 5, 7, and 9, that is, the squares of the harmonic currents at the 3rd, 5th, 7th, and 9th orders are added.

[0036] Among them, the constraint condition includes that the compensation capacitance is smaller than the remaining capacitance.

[0037] As shown in FIG. 2, in the figure, flag = 0 represents resetting the continuous update failure count to 0, flagi represents the continuous update failure count of the i-th particle, m is the failure count threshold, and if the i-th particle does not improve with respect to the objective function corresponding to the historical optimal particle even in iterations exceeding m times, it is considered that the particle may fall into a local optimal solution. At this time, the learning function mode is activated to enable the i-th particle to learn from other particles and help it escape from the local optimal solution. i represents the i-th particle in the particle set, w represents the inertia weight, k is the number of iterations, JPEG0007714817000041.jpg610 is the maximum number of iterations. JPEG0007714817000042.jpg712 and JPEG0007714817000043.jpg712 represent the minimum and maximum values of the variables of each dimension of the particle. d is the d-th data or d-dimensional data in one particle, JPEG0007714817000044.jpg612 is the maximum iteration speed, D is the total number of data in one particle, f(x i ) is the numerical value obtained by substituting the data x in the particle i and other data in the same particle into the objective function. pbest i is the historical optimal value, gbest i is the global optimal value, f(pbest i ) is the objective function value corresponding to the historical optimal value, f(gbest i ) is the objective function value corresponding to the global optimal value, and ps is the total number of particles in the particle set.

[0038] In one embodiment of the present application, as shown in FIG. 2, when the above-described constraint conditions are satisfied, analyzing the objective function corresponding to each particle in the particle set, determining the optimal solution of the objective function, and adjusting the equivalent virtual impedance of the inverter of each node based on the optimal solution is Analyze the objective function corresponding to each of the particles in the particle set, compare the objective functions corresponding to all the particles in the particle set, determine the particle corresponding to the minimum objective function, and mark the particle as the global optimal particle, including: Exemplarily, before starting the analysis of optimization, it is necessary to initialize the parameters in the flow. Assume that there are 100 particles in the particle set, and each particle has 5 pairs (one real part and one imaginary part with one equivalent virtual impedance form one pair of data), and a total of 10 data are included. Substitute these 10 data into the calculation of the objective function to calculate the value of one objective function, calculate the objective function values of 100 particles, and then find the minimum objective function value among them. The particle corresponding to it is the global optimal particle among the current 100 particles.

[0039] Input the randomly generated initial particle, initial iteration velocity, and initial optimal particle into the particle update model to update the first particle in the particle set. Among them, the particle update model is for updating the next particle based on the previous particle and the iteration velocity of the previous particle. Exemplarily, the randomly generated initial particle, initial iteration velocity, and initial optimal particle are mainly for starting the update of the first particle. This is because for the first particle, there is no such data as the previous iteration velocity, the previous particle, and the historical optimal particle before, so it needs to be randomly generated. Also, the particle update model is JPEG0007714817000045.jpg1092 JPEG0007714817000046.jpg1041 where In the formula, JPEG0007714817000047.jpg811 is the variable of the d - dimension in the (k + 1) - th iteration of the i - th particle, JPEG0007714817000048.jpg811 is the d-dimensional velocity of the i-th particle at the (k + 1)-th iteration, where 1 ≤ d ≤ D, where d means the d-th data in the particle, D represents the total amount of data in the particle, ω is the inertia weight, JPEG0007714817000049.jpg89 is the d-dimensional velocity of the i-th particle at the k-th iteration, c1 is the acceleration coefficient, r1 is a random number within the range [0, 1], JPEG0007714817000050.jpg825 is the d-th numerical value corresponding to the optimal fitness of the individual history of the fi(d)-th particle at the k-th iteration.

[0040] Also, since a single particle contains multiple data, it is necessary to update each of the multiple data in a single particle one by one. After all the data in a single particle have been updated, proceed to the next step.

[0041] Compare the objective function of the first particle after update with the objective function of the initial optimal particle, If the objective function of the first particle after update is smaller than the objective function of the initial optimal particle, then replace the initial optimal particle with the first particle after update and record it as the historical optimal particle. However, if the objective function of the first particle after update is greater than or equal to the objective function of the initial optimal particle, record the initial optimal particle as the historical optimal particle. Note that the particle after being updated by the particle update model is a new particle and is not included in the 100 particles in the particle set, which corresponds to an expansion of the particle set to increase the number of particles and find an optimal solution closer to the actual optimal particle.

[0042] Input the particle after update, the historical optimal particle, and the current iteration velocity into the particle update model to obtain the next iteration velocity and the next particle after update, and sequentially update all the particles in the particle set in a cyclic manner, Exemplarily, the update process is continuously repeated using the above particle update model, and 100 particles are updated one by one. In this update, 100 new particles are obtained, thereby enriching the particle data.

[0043] After each update, the objective function corresponding to the updated particle is compared with the historical optimal particle. If the objective function corresponding to the updated particle is smaller than the objective function corresponding to the historical optimal particle, it indicates that a new optimal particle has been found. The historical optimal particle is replaced with the updated particle, and the historical optimal particle at this time is the newly obtained particle. In this way, by comparing each time, the historical optimal solution can be obtained.

[0044] The objective function corresponding to the updated particle is compared with the objective function corresponding to the global optimal particle. Since this new particle is not the data among the existing 100 particles, by comparing the objective function value corresponding to the newly obtained particle with the objective function value of the global optimal particle, there is a possibility of finding a particle better than the original global optimal particle.

[0045] If the objective function corresponding to the updated particle is greater than or equal to the objective function corresponding to the global optimal particle, this indicates that the harmonic processing effect of the original global optimal particle is still better, and the global optimal particle is maintained as it is.

[0046] If the objective function corresponding to the updated particle is smaller than the objective function corresponding to the global optimal particle, it indicates that data better than the original global optimal particle has been found. The global optimal particle is replaced with the updated particle, and an exchange is performed on the global optimal particle. After all the particles in the particle set are updated, one iteration update of the particle set is completed. At this time, a new particle set is obtained, and the original particle set is already covered by the new particle set. Based on the new particle set, the new particle set is updated again until the particle set is updated a predetermined number of times to obtain the final global optimal particle. For example, if the predetermined number of times is 2000, that is, one global optimal particle is found from 2000 * 100 pieces of data. The global optimal particle found at this time is closer to the actual optimal particle. Thereby, with an automatic optimization method, an optimal solution with an increased processing effect is found, and the equivalent virtual impedance of each node is adjusted and controlled.

[0047] Based on the global optimal particle, the equivalent virtual impedance at each node in the microgrid is adjusted.

[0048] In this embodiment, when there is no overload, each particle learns from the historical optimal particle and the global optimal particle of the swarm simultaneously, continuously updates each particle, and maximally utilizes the remaining capacity of the inverter to achieve a better harmonic processing effect, reducing problems such as an increase in circuit loss due to harmonics, an increase in equipment loss, and an impact on the stability of the system. Each particle represents a combination of one of all the equivalent virtual impedances of the system. Among them, each dimension of each particle represents the numerical value of the virtual impedance of each inverter. By solving the constructed optimization equation and obtaining the output value of the virtual impedance of each inverter after optimization and sending it to each inverter, effective processing of the harmonic distortion rate of the microgrid system can be realized.

[0049] In one embodiment of the present application, as shown in FIG. 2, comparing the objective function corresponding to the updated particle with the historical optimal particle after each update as described above is to determine whether the value taken by the particle is within a predetermined range after each update When the value taken by the particle exceeds the predetermined range, an error occurs in the update of the particle, indicating that the particle exceeds the predetermined range, and it is meaningless to continue circulating. At this time, the cyclic update is terminated. When the value taken by the particle is within the predetermined range, it includes comparing the objective function corresponding to the updated particle with the historical optimal particle.

[0050] In this embodiment, the numerical value in the updated particle is checked once, and when it exceeds the limit, continuous optimization is avoided to reduce unnecessary analysis processes.

[0051] In one embodiment of the present application, as shown in FIG. 2, after the above-mentioned comparison of the objective function corresponding to the updated particle with the historical optimal particle after each update, the method for global optimization of the power quality of the microgrid is as follows. When the objective function corresponding to the updated particle is greater than or equal to the objective function corresponding to the historical optimal particle, mark the update of this time as a failed update, and further include counting the number of consecutive failed updates. Exemplarily, the role of counting the number of consecutive failed updates is as follows. If the fitness of a particle is not improved even in the m - time iteration process, it is considered that the particle may fall into a local optimal solution. By executing the learning function mode for it, the particle can have a chance to learn from the optimal values in the corresponding dimensions of other particles in each dimension.

[0052] When the number of consecutive failed updates is less than the failure - count threshold, continue the cyclic update. When the number of consecutive failed updates is greater than or equal to the failure - count threshold, convert the current particle to the learning function mode, obtain the particle after learning, reset the number of consecutive failed updates, replace the current particle with the particle after learning, and re - input it into the cyclic update. Exemplarily, since the constructed equation is a non-linear equation, when solving according to the previous embodiment, there is a problem that it is likely to fall into a local optimal solution. To avoid the solution result from falling into the local optimal solution as much as possible and solve this problem, when it is detected that the number of consecutive update failures is greater than or equal to the failure count threshold, at this time, it indicates that the optimization solution falls into the local optimal solution. At this time, escape from the replenishment by sequentially updating and optimizing, and insert a learning function mode that randomly optimizes backward into it to break the current situation of falling into the local optimal solution. For example, when m is equal to 5, starting from the 40th particle, when it begins to appear that the objective function value corresponding to the updated particle is greater than or equal to the objective function value corresponding to the historical optimal particle, and all five consecutive particles are like this, when it comes to the 46th particle, the particle is converted to the learning function mode.

[0053] When the objective function corresponding to the updated particle is smaller than the objective function corresponding to the historical optimal particle, reset the number of consecutive update failures.

[0054] For example, starting from the 40th particle, it begins to appear that the objective function value corresponding to the updated particle is greater than or equal to the objective function value corresponding to the historical optimal particle. However, in the 43rd data, it appears that the objective function value corresponding to the updated particle is smaller than the objective function value corresponding to the historical optimal particle. In this case, the effect of the updated particle has improved. At this time, reset the number of consecutive update failures to 0.

[0055] In one embodiment of the present application, when the number of consecutive update failures is greater than or equal to the failure count threshold as described above, converting the current particle to the learning function mode, obtaining the particle after learning, resetting the number of consecutive update failures, and replacing the current particle with the particle after learning and re-investing it into the cyclic update is When the number of consecutive update failures is greater than or equal to the failure count threshold, generate a random number, and determine whether the random number is smaller than the learning probability factor. Among them, the learning probability factor JPEG0007714817000051.jpg65 is calculated based on the position number of the current particle in the particle set and the total number of particles in the particle set.

[0056] Exemplarily, before each iteration, each particle is ranked in ascending order according to the corresponding objective function. JPEG0007714817000052.jpg630, in the formula, N represents the total number of particles. JPEG0007714817000053.jpg54 indicates that the corresponding particle corresponds to the i-th position in the ranking. The smaller i is, the better the solution result is, indicating that the probability of this particle learning from other particles is smaller and the number of particles learning is smaller. Conversely, the larger i is, the worse the solution result is, indicating that it has a larger learning probability factor. JPEG0007714817000054.jpg66, has a higher possibility of learning from other particles, and at the same time, the number of particles learning also increases. Among them, The range of JPEG0007714817000055.jpg66 is [1 / (2*N), 1 / 2].

[0057] When the random number is smaller than the learning probability factor, randomly select n particles that are behind the current particle and have not been updated in the particle set. Among them, JPEG0007714817000056.jpg568, in the formula, N represents the total number of particles. JPEG0007714817000057.jpg54 indicates that the corresponding particle corresponds to the i-th position in the ranking. [x] is the rounding function, [x] represents the largest integer not exceeding x, and the round function returns a numerical value that is the result of rounding a number to a predetermined number of decimal places.

[0058] Compare the objective functions corresponding to the selected n particles, select the particle corresponding to the minimum objective function among them, mark the particle as the learning optimal particle, and replace the current particle with the learning optimal particle. When the random number is greater than or equal to the learning probability factor, replace each piece of data in the current particle with the data that is optimal for the history of the individual of the data. Exemplarily, assume that one microgrid has five nodes, one distributed power source and one inverter are connected to each node, and one equivalent virtual impedance is connected at the position where the inverter is connected to the bus. When global optimization is not considered, for the harmonics in only one node, in order to reduce the harmonic distortion rate, on the premise that the capacity constraint condition of the inverter is satisfied, select the numerical value of the equivalent virtual impedance that minimizes the voltage harmonic distortion rate from the historical operation data of the node, and use this numerical value as the data that is optimal for the history of the individual.

[0059] Finally, after resetting the continuous update failure count, reintroduce the updated current particle into the cyclic update, let it enter the cyclic update again, and update the particles in the particle set one by one.

[0060] In this embodiment, one learning function mode is established for the particles in the population. It is considered that if the fitness of one particle is not improved even in the m - time iteration process, it may fall into a local optimal solution. By executing the learning function, each piece of data in the particle can have a chance to learn from the data that is optimal for the history of the individual in the corresponding data of other particles. By performing global control in a method with better solution performance, the possibility that the solution result falls into a local optimal solution can be effectively avoided, and the optimization performance can be improved.

[0061] As shown in FIG. 6, in one embodiment of the present application, A node admittance matrix construction module 610 configured to establish a node admittance matrix of the microgrid based on an equivalent virtual impedance of an inverter of each node of the microgrid and a local load of each node; An objective function establishment module 620 configured to construct an objective function of the microgrid based on a weight value assigned to each node in the microgrid and a voltage distortion rate of each node; A constraint condition construction module 630 configured to construct a constraint condition for harmonic modulation of each node of the microgrid based on the node admittance matrix and a remaining capacity of the inverter of each node; In response to the constraint condition being satisfied, analyzing the objective function corresponding to each particle in the particle set, determining an optimal solution of the objective function, and adjusting the equivalent virtual impedance of the inverter of each node based on the optimal solution, wherein a set of values of one of a real part and an imaginary part of the equivalent virtual impedance at all nodes in the microgrid forms one of the particles, an optimal solution search module 640; A further provided large-area optimization system for power quality of a microgrid.

[0062] The large-area optimization system for power quality of a microgrid in the embodiment of the present application has a technical effect similar to that of the large-area optimization method for power quality of the above microgrid.

[0063] In one embodiment of the present application, FIG. 3 shows a structural schematic diagram of a complex bus system of a single-phase island-type microgrid, and the parameters related to the system are as shown in Table 1.

[0064] In step 1, an equivalent circuit network of harmonics of the island-type microgrid system is established.

[0065] In step 2, a corresponding objective function is constructed such that the average voltage harmonic distortion rate of each bus is minimized. The importance of each bus is the same, that is, a1 = a2 = a3 = a4 = a5 = a6 = 1 / 6, and the objective function is JPEG0007714817000058.jpg1536 as follows.

[0066] In step 3, the constraints of the capacities of the five inverters and the harmonic power flow are established, and the virtual impedances of the five inverters are taken as decision variables.

[0067] As shown in Fig. 3, no distributed power source and inverter are connected to bus 6, which indicates that there is no equivalent virtual impedance at bus 6. Therefore, since it is not necessary to adjust the equivalent virtual impedance at bus 6 as a variable, the number of decision variables is five.

[0068] Table 1 Parameters of the single-phase islanded microgrid system JPEG0007714817000059.jpg148143

[0069] In step 4, when the constraint conditions are satisfied, an optimization solution is performed for the objective function, and the related parameters are as shown in Table 2.

[0070] Among them, since the equivalent virtual impedance of the inverter is divided into resistance and inductance, that is, the real part and the imaginary part, and each is optimized, the dimension of a single particle is 10. Among them, the variable X real representing the resistance is set in the range of [0, 20], and the variable X imag representing the inductance is set in the range of [0, -Lg].

[0071] When comparing the optimal solution obtained by the method described in this application with the optimal solution obtained by a local control strategy without an optimization algorithm (the local control strategy mimics a conventional active filter and realizes the processing of voltage harmonic distortion rate by providing a harmonic impedance path as small as possible), at the 2nd second (s), each inverter of the system adopts the local control strategy, and at the 16th second, when adopting the optimized equivalent virtual impedance value calculated by the global optimization control strategy, the processing effect of the voltage harmonic distortion rate of bus 6 is as shown in Fig. 4. In Fig. 4, since the sine curve of the voltage appears multiple times within 1 second, when the unit time length of the time axis is short, adjacent sine curves approach each other infinitely. After many sine curves approach each other infinitely, when the length of the time axis is set to the length shown in Fig. 4, the voltage change curve presents a black area as shown in Fig. 4, and it is difficult to distinguish the values of each point of the sine curve with the naked eye, so it is necessary to cut out and expand the voltage curve in a certain time period to be observed. In Fig. 4, some voltage curves before 2s, before 16s, and before 20s are cut out and expanded respectively for observing the change situation after the voltage is contaminated by harmonics or regulated and controlled.

[0072] Table 2 Parameter settings in the optimization solution algorithm JPEG0007714817000060.jpg55122

[0073] Table 3 Voltage harmonic distortion rate of each bus under different controls JPEG0007714817000061.jpg43136

[0074] When the local control strategy is adopted at the 2nd second and a stable operating state is reached, the harmonic voltage distortion rate of bus 6 is 4.18%. When the equivalent virtual impedance value after global optimization is lowered at the 16th second, the distortion rate stably becomes 3.61%. The capacity change of each inverter is as shown in Fig. 5.

[0075] The processing effects of the voltage harmonic distortion rates of each bus are as shown in Table 3. It can be seen that the global optimization control strategy can improve the voltage quality of each node in the system when the inverter does not become overloaded, compared with the local control strategy.

Claims

1. establishing a node admittance matrix of the microgrid based on an equivalent virtual impedance of an inverter of each node of the microgrid and a local load of each node; constructing an objective function of the microgrid based on a weight value assigned to each node in the microgrid and a voltage distortion rate of each node; constructing a constraint condition for harmonic modulation of each node of the microgrid based on the node admittance matrix and a remaining capacity of the inverter of each node; when the constraint condition is satisfied, analyzing the objective function corresponding to each particle in the particle set, determining an optimal solution of the objective function, adjusting the equivalent virtual impedance of the inverter of each node based on the optimal solution, and one type of value of a real part and an imaginary part of the equivalent virtual impedance at all nodes in the microgrid forms one of the particles, and the particle set includes a plurality of the particles; The constructing of the objective function of the microgrid based on the weight value assigned to each node in the microgrid and the voltage distortion rate of each node as described above includes: determining the voltage distortion rate of each node based on an effective value of a fundamental wave voltage and an effective value of a harmonic voltage of each node; multiplying and then adding the voltage distortion rate of each node and the weight value of each node in correspondence to obtain the objective function of the microgrid; A method for global optimization of power quality of a microgrid.

2. The objective function of the microgrid is: wherein N represents the number of nodes of the microgrid; is the weight value corresponding to the j-th node; is the voltage distortion rate corresponding to the j-th node; The voltage distortion rate is: wherein is the effective value of the fundamental wave voltage at the j-th node; is the effective value of the h-th harmonic voltage at the j-th node; wherein is the node admittance matrix at the h-th harmonic; is the h-th harmonic current at the j-th node. The method for global optimization of power quality of a microgrid according to Claim 1.

3. The constructing of the constraint condition for harmonic modulation of each node of the microgrid based on the node admittance matrix and the remaining capacity of the inverter of each node as described above includes: Determining the remaining capacity of the inverter based on the rated capacity of the inverter, the active power output by the inverter, and the reactive power output by the inverter; Determining the compensation capacity of the inverter based on the fundamental wave voltage of the inverter and the harmonic currents of each order flowing through the inverter, including: The constraint condition includes that the compensation capacity is smaller than the remaining capacity. The method for global optimization of power quality of a microgrid according to claim 1.

4. The remaining capacity is where is the remaining capacity of the inverter, is the rated capacity of the inverter, P is the active power output by the inverter, Q is the reactive power output by the inverter, The compensation capacity is where is the compensation capacity of the inverter, is the fundamental wave voltage of the inverter, is the harmonic current flowing through the inverter, The method for global optimization of power quality of a microgrid according to claim 3.

5. When the above-described constraint condition is satisfied, analyzing the objective function corresponding to each particle in the particle set, determining the optimal solution of the objective function, and adjusting the equivalent virtual impedance of the inverter of each node based on the optimal solution includes: Analyzing the objective function corresponding to each particle in the particle set, comparing the objective functions corresponding to all the particles in the particle set, and determining the particle corresponding to the minimum objective function as the global optimal particle; Inputting the randomly generated initial particle, initial iteration speed, and initial optimal particle into a particle update model for updating the next particle by the previous particle and the iteration speed of the previous particle, and updating the first particle in the particle set; Comparing the objective function of the updated first particle with the objective function of the initial optimal particle; In response to the objective function of the updated first particle being smaller than the objective function of the initial optimal particle, recording the updated first particle as the historical optimal particle instead of the initial optimal particle; Inputting the updated particle, the historical optimal particle, and the current iteration speed into the particle update model to obtain the next iteration speed and the next updated particle, and sequentially and cyclically updating all the particles in the particle set. After each update, comparing the objective function corresponding to the updated particle with the historical optimal particle; In response to the objective function corresponding to the updated particle being smaller than the objective function corresponding to the historical optimal particle, replacing the historical optimal particle with the updated particle; Comparing the objective function corresponding to the updated particle with the objective function corresponding to the global optimal particle; In response to the objective function corresponding to the updated particle being smaller than the objective function corresponding to the global optimal particle, replacing the global optimal particle with the updated particle; After all the particles in the particle set are updated, one iterative update of the particle set is completed, updating the particle set a predetermined number of times to obtain the final global optimal particle; Based on the global optimal particle, adjusting the equivalent virtual impedance at each node in the microgrid, including; The method for global optimization of power quality of a microgrid according to claim 1.

6. The above-mentioned comparing the objective function corresponding to the updated particle with the historical optimal particle after each update means that: After each update, determining whether the value taken by the particle is within a predetermined range; In response to the value taken by the particle exceeding the predetermined range, ending the cyclic update; In response to the value taken by the particle being within the predetermined range, comparing the objective function corresponding to the updated particle with the historical optimal particle, including; The method for global optimization of power quality of a microgrid according to claim 5.

7. After the above-mentioned comparing the objective function corresponding to the updated particle with the historical optimal particle after each update, In response to the objective function corresponding to the updated particle being greater than or equal to the objective function corresponding to the historical optimal particle, marking the update of this time as a failed update and counting the number of consecutive failed updates; In response to the number of consecutive failed updates being smaller than the failure count threshold, continuing the cyclic update; In response to the number of consecutive failed updates being greater than or equal to the failure count threshold, converting the current particle to the learning function mode, obtaining the particle after learning, resetting the number of consecutive failed updates, replacing the current particle with the particle after learning and re-entering the cyclic update; responsive to the objective function corresponding to the updated particle being smaller than the objective function corresponding to the historical optimal particle, further including resetting the continuous update failure count The method for global optimization of power quality of a microgrid according to claim 5

8. responsive to the continuous update failure count being greater than or equal to a failure count threshold, converting the current particle to a learning function mode, obtaining the particle after learning, resetting the continuous update failure count, replacing the current particle with the particle after learning, and re-investing it into cyclic update responsive to the continuous update failure count being greater than or equal to a failure count threshold, generating a random number, and determining whether the random number is smaller than a learning probability factor calculated based on the position number of the current particle in the particle set and the total number of particles in the particle set responsive to the random number being smaller than the learning probability factor, randomly selecting n unupdated particles behind the current particle in the particle set comparing the objective functions corresponding to the selected n particles, selecting the particle corresponding to the minimum objective function among them, marking the particle as the learning optimal particle, and replacing the current particle with the learning optimal particle responsive to the random number being greater than or equal to the learning probability factor, replacing each data in the current particle with the data optimal for the history of the individual of the data including resetting the continuous update failure count and then re-investing the updated current particle into cyclic update The method for global optimization of power quality of a microgrid according to claim 7

9. after comparing the objective function corresponding to the updated particle with the objective function corresponding to the global optimal particle as described above responsive to the objective function corresponding to the updated particle being greater than or equal to the objective function corresponding to the global optimal particle, further including maintaining the global optimal particle as it is The method for global optimization of power quality of a microgrid according to claim 5

Images