Wind power high-frequency resonance suppression method based on network construction type SVG virtual admittance adjustment

Abstract

The invention discloses a wind power high-frequency resonance suppression method based on network construction type SVG virtual admittance adjustment, and belongs to the technical field of wind field grid-connected converter control. The invention aims to provide a wind power high-frequency resonance suppression method based on virtual admittance adjustment of a network construction type SVG, which optimizes virtual admittance parameters of the network construction type double-loop SVG based on a particle swarm algorithm and realizes high-frequency resonance suppression of a wind power plant system. Position and speed initialization is carried out on population particles, Fourier decomposition calculation is carried out on grid-side current when high-frequency resonance occurs, an objective function value THD is obtained, an example position corresponding to the minimum value min (THD) of the objective function is selected, population internal particle sharing is carried out, iterative optimization is carried out, and finally position information corresponding to the optimal position of the particles is output. And an optimal virtual admittance value can be obtained. According to the invention, suppression of high-frequency resonance of a wind power plant system can be realized, the advantage of high dynamic response speed of a current inner loop is reserved, and overcurrent when a fault occurs can be limited.

Description

Technical Field

[0001] The invention belongs to the technical field of wind farm grid-connected converter control. Background Art

[0002] With the rapid growth of the installed capacity of new energy represented by wind power in my country, the interaction of various power electronic equipment in wind farms has led to more frequent high-frequency oscillations in the system. Resonance suppression is carried out by directly hanging SVG control in the farm. Existing resonance suppression methods such as control parameter improvement and adding additional control loops can only improve the impedance characteristics of a certain fixed frequency band. When the operation of the power grid changes and causes a large range of changes in the resonant frequency or broadband oscillations, the control effects of the above methods are not ideal.

[0003] To address the problems of system resonance and instability caused by the interaction between large-scale grid-following control converters and weak power grids, the grid-following SVG within the wind farm is transformed into a grid-forming type. This can not only maintain the original dynamic compensation reactive power capability, but also provide additional inertial support and damping for the system, effectively improving the stability of the wind farm system. Summary of the invention

[0004] The purpose of the present invention is to optimize the virtual admittance parameters of a meshed double-loop SVG based on a particle swarm algorithm, and to realize a wind power high-frequency resonance suppression method based on meshed SVG virtual admittance regulation to suppress high-frequency resonance of a wind farm system.

[0005] The steps of the present invention are: S1. Obtain the initial value of virtual admittance according to the rated voltage and power of the grid-type SVG converter: Where: S S and U S are the rated power and rated phase voltage of the grid-forming SVG respectively; G v0 and B v0 are the initial values ​​of virtual conductance and susceptance, respectively; The grid-side current harmonic distortion rate min (THD) is selected as the objective function: Where: h1 is the fundamental frequency component energy, h2, h3…h n is the energy of each harmonic component; S2. Initialize and constrain key system operation parameters and links: S21, PCC voltage U PCC The range constraints are: S22. The power factor cosλ at the PCC point of the wind farm outlet should meet the following requirements: cosλmin≤cosλ≤1 (4) S23, wind speed V w Section and fan output P w The range is: S24, grid-type SVG power angle θ S The range is: S25, SVG output current I S Amplitude constraint: In formulas (3) to (7): and are the lower and upper limits of the 35kV PCC voltage of the wind farm respectively; cosλ min is the minimum power factor at the PCC point; and They are the minimum and maximum wind speeds at which the wind turbine can be connected to the grid and operated; and are the minimum and maximum output of the fan respectively; and They are the minimum and maximum stable power angles of the mesh-type SVG respectively; and They are the minimum and maximum values ​​of the SVG output current respectively; S3. Important properties of particles include position speed The iteration formula is as follows: Where: ω is the inertia weight factor; t is the number of iterations; c1 and c2 are learning factors respectively; is the optimal position of individual particles; is the global optimal position; is the position of the particle at the tth iteration; r1 and r2 are both random numbers uniformly distributed in the interval [0-1]; assuming that the number of particles in the population is n, the position information of the i-th particle is encoded as: Where: s is the number of wind turbines; j is the number of spatial dimensions; After the iteration is completed, the position information x corresponding to the optimal position of the particle is output, that is, the optimal virtual admittance value is obtained.

[0006] The present invention can suppress high-frequency resonance of a wind farm system, which not only retains the advantage of fast dynamic response speed of the current inner loop, but also can limit overcurrent when a fault occurs. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 A diagram of a mesh-type SVG control structure provided in an embodiment of the present invention; Figure 2 A flowchart of a process for solving virtual admittance optimization using a particle swarm algorithm provided in an embodiment of the present invention; Figure 3 A grid diagram of a double-fed wind farm system including SVG provided in an embodiment of the present invention; Figure 4 It is a simulated waveform diagram of grid-side voltage and current after the wind farm system is put into an unloaded line when the grid-type SVG is not optimized for virtual admittance provided in an embodiment of the present invention; Figure 5 It is a Fourier decomposition diagram of the grid-side current when the grid-type SVG is not optimized for virtual admittance and the wind farm system resonance occurs, provided in an embodiment of the present invention; Figure 6 It is a simulated waveform diagram of grid-side voltage and current after the wind farm system is put into an unloaded line when the grid-type SVG adopts virtual admittance optimization provided in an embodiment of the present invention; Figure 7 When the mesh type SVG adopts virtual admittance optimization, the Figure 5 Fourier decomposition diagram of the grid-side current at the corresponding moment. DETAILED DESCRIPTION

[0008] From the perspective of control structure, meshed SVG can be divided into two categories: single-loop control and dual-loop control. Since meshed SVG dual-loop control adds an inner potential control loop and a current inner loop compared to single-loop control, the output current of meshed SVG can be monitored in real time. Therefore, the optimization of the virtual admittance parameters of meshed dual-loop SVG based on the particle swarm algorithm can suppress the high-frequency resonance of the wind farm system, which not only retains the advantage of fast dynamic response speed of the current inner loop, but also limits the overcurrent when a fault occurs.

[0009] In order to solve the high-frequency resonance problem of the wind farm system, limit the harm of the overcurrent output of the grid-type SVG to the equipment and system, and improve the stability of the wind farm system, the present invention provides a method for suppressing the high-frequency resonance of the wind farm system based on the virtual admittance regulation of the grid-type SVG using a particle swarm algorithm, including: The initial values ​​of virtual conductance and virtual susceptance are set according to the rated voltage and power of the grid-type SVG converter.

[0010] A particle swarm algorithm based virtual admittance parameter optimization algorithm for meshed SVG is proposed and used to select the optimal parameters of virtual admittance for meshed SVG. The specific embedding position is as follows: Figure 1 Medium G v +jB v shown.

[0011] The mesh-type SVG uses the iterated virtual admittance to reshape the impedance characteristics of the mesh-type SVG, so that the mesh-type SVG can solve the high-frequency resonance problem of the wind farm system without sacrificing the reactive power compensation capability.

[0012] The present invention uses a particle swarm algorithm to optimize and adjust the virtual admittance parameters in SVG, thereby reshaping the overall admittance of the wind farm and suppressing the high-frequency resonance of the wind farm system. The method is specifically as follows: The initial value of virtual admittance is calculated according to the rated voltage and power of the grid-type SVG converter, and the initial value of virtual admittance is calculated according to formula (1): Where: S S and U S are the rated power and rated phase voltage of the grid-forming SVG respectively; G v0 and B v0 are the initial values ​​of virtual conductance and susceptance, respectively. When the initial value of virtual admittance is applied to the grid-type SVG, when the wind farm system is disturbed (such as wind turbine switching), the harmonic content in the grid-side current of the system is large. Therefore, it is necessary to use the particle swarm algorithm to optimize the virtual admittance parameters.

[0013] From the perspective of suppressing high-frequency resonance of the wind farm system, the present invention selects the grid-side current harmonic distortion rate min (THD) as the objective function: In formula (2), h1 is the fundamental frequency component energy, h2, h3…h n is the energy of each harmonic component.

[0014] In order to ensure the stable operation of the wind farm system and the smooth optimization of the virtual admittance parameters, the key parameters and links of the system operation should be initialized and constrained. The specific constraints are as follows: 1. Wind farm 35kV PCC voltage U under continuous and stable system operation PCC The range constraints are:

[0015] 2. The power factor cosλ of the PCC point at the wind farm outlet should meet the following requirements: cosλ min ≤cosλ≤1 (4)

[0016] Wind speed V that allows wind turbine to operate normally in grid connection w Interval and corresponding fan output P w The range is:

[0017] The grid-type SVG power angle θ that can meet the stable operation of the wind farm system S The range is:

[0018] SVG output current I S Amplitude constraint: In the above formula (3-7): and are the lower and upper limits of the 35kV PCC voltage of the wind farm respectively; cosλ min is the minimum power factor at the PCC point; and They are the minimum and maximum wind speeds at which the wind turbine can be connected to the grid and operated; and are the minimum and maximum output of the fan respectively; and They are the minimum and maximum stable power angles of the mesh-type SVG respectively; and They are the minimum and maximum values ​​of the SVG output current respectively.

[0019] Each particle a in the particle swarm algorithm can be regarded as a feasible solution that minimizes the THD value of the grid-side current of the wind farm system, and searches for the optimal position in the maximum D-dimensional space. The important properties of particles include position speed According to the best position shared by particles in the population, the properties are optimized. The iterative formula is as follows: Where: ω is the inertia weight factor; t is the number of iterations; c1 and c2 are learning factors respectively; is the optimal position of individual particles; is the global optimal position; is the position of the particle at the tth iteration; r1 and r2 are both random numbers uniformly distributed in the interval [0-1].

[0020] Assuming that the number of particles in the population is n, the position information encoding of the i-th particle is as follows: Where: s is the number of wind turbines; j is the number of spatial dimensions.

[0021] Within the variable constraint range, the position and velocity of the population particles are initialized, the wind farm system operates using the electrical parameters in the particles, and then the grid-side current is Fourier decomposed and calculated when high-frequency resonance occurs to obtain the objective function value THD. Compare the objective function values ​​of all particles, select the example position corresponding to the minimum value of the objective function min(THD), share the particles within the population, and perform iterative optimization. After the iterative operation is completed, the position information x corresponding to the optimal position of the particle is output, and the optimal virtual admittance value can be obtained. Applying this virtual admittance parameter value will change the impedance characteristics of the grid-type SVG, and then change the overall impedance characteristics of the wind farm system, so as to obtain the optimal wind farm system resonance suppression effect and improve system stability.

[0022] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention: At present, the main problems faced by SVG wind farm systems in high-frequency resonance suppression are: 1. Although existing resonance suppression strategies such as improving control parameters, adding additional control loops and installing additional control devices can suppress resonance and the equipment modification cost is low, the above suppression strategies can only improve the impedance characteristics of a certain fixed frequency band. When the operation of the power grid changes and causes the resonance frequency to change over a large range or broadband oscillation occurs, the control effect of the above methods is not ideal.

[0023] 2. Although the single-loop control of the grid-type SVG is effective in suppressing high-frequency resonance of the wind farm system, the single-loop control lacks the internal potential control loop and the current internal loop compared to the dual-loop control, which makes its own control loop bandwidth low and the dynamic control performance poor. It is also unable to limit fault overcurrent, which puts the equipment at risk of being disconnected from the grid due to overcurrent.

[0024] In view of this, the present invention provides a method for optimizing virtual admittance parameters of SVG based on particle swarm algorithm to suppress high-frequency resonance of wind farm system.

[0025] Combined with Figure 2 The method of the present invention comprises in detail: Process 1: Combine with Figure 3 The initial values ​​of virtual conductance and virtual susceptance G are calculated from the rated voltage and power of the grid-type SVG converter in the system architecture. v0 and B v0 .

[0026] Process 2: Read in wind farm system related parameters and constraints, mainly including: Wind farm system PCC voltage U PCC , PCC point power factor cosλ, wind speed of each wind turbine V w , the output of each wind turbine generator P w , SVG output current I S , grid-type SVG power angle θS , G v and B v They are the virtual conductance and susceptance values ​​of the grid-type SVG, and the electrical and control parameters of the wind farm system and the grid-type SVG in Appendix 1 and 2.

[0027] Table 1

[0028] Table 2

[0029] Process 3: Set the particle swarm algorithm population size N, the maximum number of algorithm iterations M, the learning factors c1 and c2, ω as the inertia weight factor, the mutation rate μ and other parameters.

[0030] Process 4: Initialization of all particles in the population. According to the constraint range of the key parameters of the system (see formula 3-7), the initial particle population is randomly generated, and each particle is assigned corresponding speed and position information.

[0031] Process 5: The wind farm system operates using the electrical parameters in the particles. When the wind farm system resonates, the grid-side current is Fourier decomposed to obtain the corresponding objective function value THD, and then the optimal position of the population is obtained by comparing the objective function values ​​of all particles.

[0032] Process 6: Update the speed and position information of each particle according to formula (8), adjust the new best individual position and generate a new population.

[0033] Process 7: Determine whether the algorithm has reached the maximum number of iterations M. If it is satisfied, stop the iteration, otherwise go to process 5. When the algorithm terminates, the position information x corresponding to the optimal position of the particle is output, and the optimal virtual admittance value can be obtained.

[0034] Process 8: The grid-type SVG adopts the optimal virtual admittance parameters to reshape the system impedance characteristics, thereby suppressing the high-frequency resonance of the wind farm system.

[0035] Combine the following Figure 1 The specific embodiments are as follows to introduce the technical solution of the present invention in detail: Step 1: According to the attached Figure 3 The initial value of the virtual admittance calculated by the rated voltage and power of the grid-connected SVG converter in the wind farm system architecture is shown in formula (1): Where: S S and U S are the rated power and rated phase voltage of the grid-forming SVG respectively; G v0 and B v0are the initial values ​​of virtual conductance and susceptance, respectively.

[0036] Step 2: Establish constraint conditions according to relevant national standards and State Grid corporate standards: Wind farm PCC voltage U under continuous operation PCC range (90% - 110%), the power factor cosλ cannot exceed the lower limit of 0.95, the fan speed V w Range (5m / s—25m / s), fan output P w Constraint (0-1.5MW), grid-type SVG power angle θ S In the stable region (2-20%), the wind farm output harmonic distortion rate THD≤5%, and the electrical parameters of the wind farm system and the grid-type SVG are initialized and set according to Appendix 1 and 2.

[0037] Step 3: Set the particle swarm algorithm population size N = 100, the maximum number of algorithm iterations M = 200, the learning factors c1 = 0.2 and c2 = 0.4, the inertia weight factor ω = 0.9 (start) ω = 0.001 (end), and the mutation rate μ = 0.1.

[0038] Step 4: According to formula (3-7), randomly generate the initial particle population and assign corresponding speed and position information to each particle.

[0039] Step 5: The wind farm system is operated using the electrical parameters in the particles. The grid-side current is collected when the wind farm system resonates, and Fourier decomposition is performed to obtain the corresponding objective function THD value. Then, the objective function values ​​of all particles are compared and the current optimal individual and global optimal positions of the particles are determined.

[0040] Step 6: Update the speed and position information of each particle according to formula (8), adjust the new best individual position and generate a new population.

[0041] Step 7: Determine whether the algorithm has reached the maximum number of iterations M. If so, stop the iteration, otherwise go to process 5. When the algorithm terminates, output the position information x corresponding to the optimal position of the particle, and the optimal virtual admittance value can be obtained.

[0042] Step 8: The grid-type SVG adopts the optimal virtual admittance parameters to solve the high-frequency resonance problem of the wind farm system without sacrificing the reactive power compensation capability.

[0043] According to the attached Figure 3The wind farm system structure shown in the figure is used to build a simulation model of a doubly-fed wind farm system including a grid-type SVG. Combined with the wind farm system electrical parameters and grid-type SVG control parameters provided in Tables 1 and 2, simulation verification is carried out. The verification conclusion is as follows: When the grid-side current harmonic distortion rate is detected to exceed the set value (THD>10%), the grid-type SVG virtual admittance parameter optimization algorithm is started and the virtual admittance parameters are optimized. When the wind power system has an unloaded line disturbance, the grid-type SVG adopts the initial value of the virtual admittance. The grid-side voltage and current simulation waveforms of the wind farm system are shown in the attached figure. Figure 4 Analysis Figure 4 It can be found that the system can operate normally before the disturbance occurs. After the no-load line is put into operation for 3s, the voltage and current on the grid side of the system both have harmonic distortion. The Fourier decomposition of the grid side current after the no-load line is put into operation is further performed. The results are as follows: Figure 5 Observation Figure 5 It can be seen that the harmonic components of the wind farm system resonance are mainly concentrated near the frequency of 1120Hz, and the amplitude HR at this frequency is 59.4%. In comparison, the grid-type SVG virtual admittance parameter optimization algorithm based on the particle swarm algorithm is started, and the grid-type SVG adopts the optimized virtual admittance parameters. The grid-side voltage and current simulation waveforms of the wind farm system are shown in the attached figure. Figure 6 After the no-load line is put into operation at 3s, the system can still operate stably regardless of the disturbance. The Fourier decomposition of the grid-side current after the no-load line is further performed, and the results are shown in Figure 7 Comparing the grid-side current of the wind farm before and after virtual admittance optimization, the THD is reduced from 99.44% to 0.82%, which proves the effectiveness of the resonance suppression method proposed in the present invention.

Claims

1. A method for suppressing high-frequency resonance of wind power based on virtual admittance regulation of meshed SVG, characterized by: The steps are: S1. Obtain the initial value of virtual admittance according to the rated voltage and power of the grid-type SVG converter: Where: S S and U S are the rated power and rated phase voltage of the grid-forming SVG respectively; G v0 and B v0 are the initial values ​​of virtual conductance and susceptance, respectively; The grid-side current harmonic distortion rate min (THD) is selected as the objective function: Where: h1 is the fundamental frequency component energy, h2, h3…h n is the energy of each harmonic component; S2. Initialize and constrain key system operation parameters and links: S21, PCC voltage U PCC The range constraints are: S22. The power factor cosλ at the PCC point of the wind farm outlet should meet the following requirements: cosλmin≤cosλ≤1 (4) S23, wind speed V w Section and fan output P w The range is: S24, grid-type SVG power angle θ S The range is: S25, SVG output current I S Amplitude constraint: In formulas (3) to (7): and are the lower and upper limits of the 35kV PCC voltage of the wind farm respectively; cosλ min is the minimum power factor at PCC point; and They are the minimum and maximum wind speeds at which the wind turbine can be connected to the grid and operated; and are the minimum and maximum output of the fan respectively; and They are the minimum and maximum stable power angles of the mesh-type SVG respectively; and They are the minimum and maximum values ​​of the SVG output current respectively; S3. Important properties of particles include position speed The iteration formula is as follows: Where: ω is the inertia weight factor; t is the number of iterations; c1 and c2 are learning factors respectively; is the optimal position of individual particles; is the global optimal position; is the position of the particle at the tth iteration; r1 and r2 are both random numbers uniformly distributed in the interval [0-1]; Assuming that the number of particles in the population is n, the position information of the i-th particle is encoded as: Where: s is the number of wind turbines; j is the number of spatial dimensions; After the iteration is completed, the position information x corresponding to the optimal position of the particle is output, that is, the optimal virtual admittance value is obtained.

Images