A converter electromagnetic transient simulation parallel optimization method, device and storage medium

Abstract

The application discloses a kind of converter electromagnetic transient simulation parallel optimization method, equipment and storage medium, belong to power system electromagnetic transient simulation technical field, by merging significantly reduces unnecessary intermediate variable, shortens simulation calculation process;And since C, D matrix and Un, k matrix multiplication operation with Itotal of A, B matrix and multiplication are carried out simultaneously, and the calculation process between row and row is independent, increase the parallel degree of electromagnetic transient simulation calculation;Since parameter matrix does not change with the change of switch state, it is calculated and stored in advance in simulation initialization stage, only needs to consume register reading operation time in the calculation of single simulation step;Coefficient matrix A, B, C, D are all square matrix of scale Nb*Nb, the formula for calculating Ub and Ib is the same in structure, to lay the algorithm foundation for saving logic resources while improving the calculation efficiency of simulation program using pipeline and other hardware development techniques.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of electromagnetic transient simulation of power systems, and particularly relates to a converter electromagnetic transient simulation parallel optimization method, device and storage medium. BACKGROUND

[0002] With the continuous development of current technology, the main bottom processor of commercial simulation platform has experienced the transformation from DSP, RISC to multi-core CPU and CPU cluster technology, and has more and more powerful computing power, but due to the limitations of hardware cost and performance, and the bottleneck of communication time between multiple processors, it cannot fully meet the increasing demand for high-precision small-step real-time simulation; GPU and FPGA with stronger parallel computing power are applied to hardware accelerated real-time simulation of research institutions such as colleges and commercial simulation platforms such as RTDS, however, the traditional electromagnetic transient simulation calculation program is designed for general-purpose processors based on serial logic work, and when it is directly transplanted to the processing unit with stronger parallel computing performance such as GPU and FPGA, it is difficult to fully exert its parallel performance advantage; therefore, the present application provides a converter electromagnetic transient simulation parallel optimization method, device and storage medium. SUMMARY

[0003] In view of the deficiencies of the prior art, the purpose of the present application is to provide a converter electromagnetic transient simulation parallel optimization method, which solves the technical problems mentioned in the background art.

[0004] The purpose of the present application can be achieved by the following technical solutions: a converter electromagnetic transient simulation parallel optimization method, the method comprising the following steps:

[0005] According to the electromagnetic transient simulation theory, the characteristic equation of the system element is differentiated, and an electromagnetic transient simulation model of the power network based on the node analysis method is established, and the parameterized constant admittance model is used for modeling the power electronic switching element in the network;

[0006] Based on the electromagnetic transient simulation model of the power network and the electromagnetic transient simulation theory, the electromagnetic transient simulation calculation is carried out, and the electromagnetic transient simulation calculation process based on the node analysis method in the network is represented in the form of a matrix;

[0007] The electromagnetic transient simulation calculation process based on the node analysis method represented in the form of a matrix is merged and simplified.

[0008] Preferably, the parameterized constant admittance model is selected as the mathematical model for simulation calculation of the power electronic switching element in the power system, and the basic formula is:

[0009]

[0010] Where u(t), i(t) are the voltage and current of the branch at the current time; u(t-Δt), i(t-Δt) are the voltage and current of the branch at the previous time; I his (t) is the equivalent history current source; α, β are the voltage coefficient and current coefficient of the equivalent current source; G eq represents the equivalent conductance matrix of the system.

[0011] Preferably, the parameterized constant admittance mathematical model for the power electronic switching element in the equivalent network adopts a cross initialization algorithm to reduce transient errors in the switching state switching process.

[0012] Preferably, the method for reducing transient errors in the switching state switching process is: when there are pairs of switching groups with mutually exclusive states in the converter model, a cross initialization algorithm is adopted to reinitialize the electrical state quantity of the opposite side switch of the switching group at the moment of switching state change as the basis for calculating the history current source of the current side switch, thereby further reducing the transient errors caused by harmonic and virtual power loss in the constant admittance model, and obtaining a simulation result closer to the ideal switch.

[0013] Preferably, the process of the electromagnetic transient simulation calculation process based on the node analysis method in the matrix form representing the network of the high proportion of power electronic equipment includes the following steps:

[0014] Calculate the equivalent current source of each branch

[0015]

[0016] Where the superscripts "n+1" and "n" represent the next time to be solved and the known previous time, respectively, I his , U b , I b represent the vectors composed of the equivalent current source of each branch, the voltage across the branch, and the current flowing through the branch, and the size is the total number of branches N b ; α, β are diagonal matrices with a size of N b *N b If the i-th branch corresponds to a switching element in the converter model, the i-th diagonal element α(i), β(i) in the diagonal matrix α, β is determined by the switching state and the following formula;

[0017]

[0018] Where α on , β on and α off , β offLet α(i) and β(i) represent the values ​​of the elements corresponding to the i-th branch in the α and β matrices when the switch is on or off, respectively. If the i-th branch corresponds to passive components such as inductors and capacitors, the values ​​of α(i) and β(i) are obtained by nodal analysis.

[0019] G b This is a diagonal matrix with diagonal elements representing the admittances of each branch, and has a size of N. b *N b C1 and C2 are of size N b *N b The cross-initialization matrices, C1 and C2, are diagonal matrices where all switch states in the next time step are the same as in the previous time step. If the switch state corresponding to the i-th branch changes in the next time step, let another switch forming the same bridge arm as the switch in the i-th branch be located in the j-th branch. According to the definition of cross-initialization:

[0020]

[0021] Calculate the injection current at the unknown node

[0022]

[0023] In the formula I inj,u Let N be the vector of injected currents at all nodes with unknown voltages, and its size is the number of unknown nodes. u I s The current vector injected into the node by the external power source has a magnitude of N. b M u The system correlation matrix for unknown nodes has a size of N. u *N b For any branch i and node j, M u The value is defined as:

[0024]

[0025] Solving for the voltage at the unknown node

[0026]

[0027] In the formula U n,u Let N be a vector of voltages at unknown nodes. u V n,k Let N be the vector of voltages at known nodes, with a size equal to the number of known nodes. k G uu and G uk The size of the unknown nodes is N. u *N u The self-admittance matrix and the size of the unknown nodes and known nodes are N.u *N k The mutual admittance matrix is part of the system admittance matrix;

[0028] Solving the voltage and current of each branch

[0029]

[0030]

[0031] where U n,k is the voltage vector of the known nodes, which is expanded to N b .

[0032] Preferably, the diagonal elements of the diagonal matrix depend on the branch type and system parameters.

[0033] Preferably, the merging and simplifying process includes three parts, the first part is the parameter matrix determined by system parameters and topology, including the incidence matrix M u , the system admittance matrix G b , G uu and G uk , which remains unchanged during the simulation calculation; the second part is the input matrix determined by the signals input to the system, including the switch state S, the cross initialization matrix C1, C2 reflecting the change of the switch state, the external current source I s and the voltage vector V n,k of the known nodes, reflecting the influence of the external power supply and the control system on the solution of the system state; the third part is various state quantities and intermediate variables of the system, including the branch equivalent current source I his , the branch voltage U b and the branch current I b , the voltage U n,u of the unknown nodes and the injected current I inj,u , the intermediate variables in the third part are eliminated, and other variables are represented by the branch voltage U b and the branch current I b .

[0034] Preferably, the merging and simplifying calculation process includes the following steps:

[0035] The simultaneous equations (1), (2), (3) and (4) are obtained:

[0036]

[0037] Without affecting the calculation results, the size of G uk and V n,k is adjusted to N u *N bwith N b then

[0038] Similarly, by combining equations (1), (2), (5), and (6), we can obtain

[0039]

[0040] Equations (6) and (7) can be expressed as

[0041]

[0042] wherein,

[0043]

[0044] wherein A, B, C, and D are coefficient matrices of size N b *N b , and define b C1U b + βC2I b + I s as the total injection current I total of the system. Thus, multiple serially executed calculation steps are converted into matrix multiplication calculations between rows, which provides an algorithmic basis for exploiting the advantages of parallel computing performance.

[0045] Advantages of the present application:

[0046] Compared with the conventional electromagnetic transient simulation calculation method using a serial process, the present application significantly reduces unnecessary intermediate variables by merging, shortening the simulation calculation process. Moreover, the matrix multiplication operation of the C and D matrices with Un,k can be performed simultaneously with the multiplication of the A and B matrices with Itotal, and the calculation processes between rows are independent of each other, thereby increasing the parallel degree of electromagnetic transient simulation calculation. Since the parameter matrices do not change with the switch state, they can be calculated and stored in advance during the simulation initialization stage, and only the time for register reading operation is consumed during the calculation of a single simulation step. In the formula, the coefficient matrices A, B, C, and D are all square matrices of size Nb*Nb, and the formulas for calculating Ub and Ib are completely identical in structure, thereby laying an algorithmic foundation for improving the simulation program calculation efficiency while saving logic resources by using hardware development techniques such as pipelining. BRIEF DESCRIPTION OF DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, for those of ordinary skill in the art, without paying creative labor, other drawings can also be obtained based on these drawings.

[0048] Figure 1 Flow chart of the present application;

[0049] Figure 2 Flow chart of the present application;

[0050] Figure 3 Flow chart of the present application;

[0051] Figure 4 Flow chart of the present application;

[0052] Figure 5 Comparison chart of the simulation method of the present application and PSCAD simulation of the total active power on the AC side of the multi-converter;

[0053] Figure 6 Comparison chart of the simulation method of the present application and PSCAD simulation of the C-phase voltage on the parallel bus;

[0054] Figure 7 Comparison chart of the simulation method of the present application and PSCAD simulation of the total output current on the parallel bus B-phase;

[0055] Figure 8 Comparison chart of the simulation method of the present application and PSCAD simulation of the input voltage on the DC side of the chopper of the converter 1. DETAILED DESCRIPTION

[0056] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative labor fall within the protection scope of the present application.

[0057] As shown in Figures 1-3 A parallel optimization method for converter electromagnetic transient simulation, comprising the following steps:

[0058] (1) Based on the electromagnetic transient simulation theory, the characteristic equation of the system element is differentiated to establish a network electromagnetic transient simulation model based on the node analysis method, wherein the parameterized constant admittance model is used for modeling the switching element, and the cross initialization algorithm is used to reduce the transient error in the switching process of the switching state;

[0059] (2) The electromagnetic transient simulation calculation process based on the node analysis method in the network of high proportion of power electronic equipment is represented in matrix form.

[0060] (3) The electromagnetic transient simulation calculation process represented in matrix form is combined and simplified.

[0061] It needs to be further explained that in step (1), the model of power electronic switch can be expressed as:

[0062]

[0063] In the formula, u(t), i(t) are voltage and current of branch at solving time; u(t-Δt), i(t-Δt) are voltage and current of branch at previous time; I his (t) is equivalent historical current source; α, β are voltage and current coefficients of equivalent current source. When there are pairs of switches in the converter model that are mutually exclusive in state, the cross initialization algorithm can be used to further reduce the error caused by harmonic and virtual power loss in the constant admittance model, and a simulation result closer to the ideal switch is obtained.

[0064]

[0065] It needs to be further explained that in step (2), according to the electromagnetic transient simulation theory, each flow in the simulation calculation can be expressed in matrix form as follows:

[0066] (21) Calculate the equivalent current source of each branch

[0067]

[0068] In the formula, superscripts “n+1” and “n” represent the next time to be solved and the known previous time, respectively, I his , U b , I b represent the vectors composed of the equivalent current source of each branch, the voltage across the branch and the current flowing through the branch, and the size is N b . α, β are diagonal matrices with a size of N b *N b , wherein the values of diagonal elements depend on the type of branch and system parameters. If the i-th branch corresponds to a switch element in the converter model, the values of α(i), β(i) are determined by the switch state and the following formula;

[0069]

[0070] In the formula, α on , β on and α off , β off respectively represent the values of the elements corresponding to the i-th branch in the α, β matrix when the switch is on or off. If the i-th branch corresponds to a passive element such as inductor, capacitor, etc., the values of α(i), β(i) can be obtained according to the node analysis method. G b is a diagonal matrix with diagonal elements being the admittance of each branch, and the size is N b *Nb C1, C2 are diagonal matrices with size N b *N b . If the switch state of the i-th branch changes in the next time step, and the other switch in the same bridge leg as the i-th branch is located in the j-th branch, according to the definition of cross initialization:

[0071]

[0072] (22) Calculate the injection current of the unknown node

[0073]

[0074] where I inj,u is the injection current vector of all voltage unknown nodes, with size N u unknown nodes, I s is the current vector injected by the external power supply to the node, with size N b . M u is the system incidence matrix of the unknown nodes, with size N u *N b . For any branch i and node j, the value of M u is defined as:

[0075]

[0076] (23) Solve the voltage of the unknown node

[0077]

[0078] where U n,u is the voltage vector of the unknown nodes, with size N u ; V n,k is the voltage vector of the known nodes, with size N k known nodes. G uu and G uk are the self- admittance matrix of the unknown nodes (size N u *N u ) and the mutual admittance matrix between the unknown nodes and the known nodes (size N u *N k ), respectively, which are part of the system admittance matrix.

[0079] (24) Solve the voltage and current of each branch

[0080]

[0081]

[0082] In the formula U n,k To expand the size to N b The voltage vector of a known node.

[0083] It needs to be further explained that, in the specific implementation process, step (3) involves merging and simplifying the above serial calculation process. In all stages of solving the system state, each matrix can be divided into three parts: the first part is the parameter matrix determined by the system parameters and topology, including the correlation matrix M. u System admittance matrix G b G uu and G uk The first part of the matrix remains unchanged throughout the simulation. The second part is the input matrix determined by the signals from the external input system, including the switching state S, the cross-initialization matrices C1 and C2 reflecting the changes in the switching state, and the externally injected current source I. s and the voltage vector V of the known node n,k The first part reflects the influence of external power sources and control systems on solving for the system state. The third part consists of various state variables and intermediate variables of the system, including the equivalent current source I of the branches. his Branch voltage U b and branch current I b Voltage U of unknown node n,u and injection current I inj,u Wait. Consider eliminating the intermediate variables in the third part, and using the branch voltage U. b and branch current I b To represent other variables, the specific steps are as follows:

[0084] (31) Combining the formulas in steps (21), (22), and (23), we can obtain the following:

[0085]

[0086] (32) Substituting the formula in step (31) into the formula in step (24) yields the following result.

[0087]

[0088] (33) Without affecting the calculation results, G is adjusted by adding zero elements. uk and V n,k The size is adjusted to N u *N b With N b Then the formula in step (32) can be expressed as

[0089]

[0090] (34)The formulas in the simultaneous electromagnetic transient simulation calculation process can be obtained as follows

[0091]

[0092] (35)The formulas in steps (33) and (34) are expressed as follows

[0093]

[0094] wherein,

[0095]

[0096] wherein A, B, C, and D are coefficient matrices of scale N b *N b , and αG b C1U b +βC2I b +I s is the comprehensive injection current I total of the system. Thus, multiple serial calculation steps are converted into matrix multiplication calculations between rows, which provides an algorithm basis for taking advantage of parallel computing performance.

[0097] In order to verify the effectiveness of the above method, a bipolar photovoltaic multi-inverter parallel grid model in Figure 4 is taken as an example, and the above method is used to build an electromagnetic transient simulation model of the network electrical system in an FPGA hardware simulation platform, and the same model with the same simulation step is established in PSCAD / EMTDC as a control. The electrical system simulation step is set to 2 μs, and the system parameters are shown in the following table:

[0098] Table 1 Parameters of the two-stage photovoltaic multi-inverter parallel grid test example

[0099]

[0100] The simulation scene is set as follows: the inverter control system adopts constant power control, wherein the active power reference value is initially set to 50 kW, and the reactive power reference value is initially set to 0 kvar. At 2 s, the active power reference value is adjusted to 55 kW, and the DC side power voltage is raised from 390 V to 420 V. The same simulation model with a step of 2 μs is established in PSCAD / EMTDC as a control, and the simulation results are shown in Figure 5 , which are basically consistent with PSCAD / EMTDC, proving the accuracy of the above simulation method.

[0101] In the above embodiment, the clock frequency of the FPGA internal simulation calculation module is 142.8 MHz, and the average time consumption of simulation calculation per step is 1.084 μs, which is lower than the simulation step of 2 μs. Benefiting from the strong parallel computing performance of the method, the above simulation embodiment has the computing power conditions to realize the real-time simulation of the 2 μs level small step.

[0102] In the description of the present specification, the description referring to the terms "one embodiment", "an example", "a specific example" and the like means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0103] The basic principles, main features and advantages of the present application are shown and described above. It should be understood by those skilled in the art that the present application is not limited by the above embodiments, and the above embodiments and descriptions in the specification are only to illustrate the principles of the present application. Without departing from the spirit and scope of the present application, various changes and improvements can be made to the present application, and these changes and improvements all fall within the scope of the present application.

Claims

1. A parallel optimization method for electromagnetic transient simulation of a converter, characterized in that, The method includes the following steps: Based on electromagnetic transient simulation theory, the characteristic equations of system components are differentiated, and an electromagnetic transient simulation model of power network based on nodal analysis method is established. For power electronic switching components in the network, a parameterized constant admittance model is used for modeling. Electromagnetic transient simulation calculations are carried out based on the electromagnetic transient simulation model and electromagnetic transient simulation theory of power networks. The electromagnetic transient simulation calculation process based on the node analysis method in the network is represented in matrix form. The electromagnetic transient simulation calculation process based on the node analysis method, represented in matrix form, is merged and simplified. The merging and simplification process consists of three parts. The first part is a parameter matrix determined by system parameters and topology, including the correlation matrix. M u System admittance matrix G b , G uu and G uk This part of the matrix remains unchanged throughout the simulation calculation; The second part is the input matrix determined by the signals from the external input system, including the switching states. S Cross-initialization matrix reflecting the changes in switch states C 1. C 2. External current injection source I s and the voltage vector of the known node V n,k The first part reflects the influence of external power sources and control systems on solving for the system state; the second part contains various state variables and intermediate variables of the system, including equivalent current sources in branches. I his Branch voltage U b and branch current I b Voltage of unknown nodes U n,u and injected current I inj,u Eliminate the intermediate variables in the third part and use the branch voltage. U b and branch current I b Indicates other variables; The simplified calculation process includes the following steps: Combining formulas (1), (2), (3), and (4), we get: Without affecting the calculation results, add zero elements to... G uk and V n,k The scale was adjusted to N u N b and N b ,but (6) Similarly, by combining formulas (1), (2), (5), and (6), we can obtain... （7） Formulas (6) and (7) can be expressed as: in, In the formula A , B , C , D For scale N b N b The coefficient matrix is ​​defined. αG b C 1 U b +βC 2 I b + I s Injecting the overall current into the system I total Thus, multiple consecutively executed serial computation steps are transformed into matrix multiplication calculations that are independent of each other between rows, providing an algorithmic basis for leveraging the performance advantages of parallel computing.

2. The parallel optimization method for electromagnetic transient simulation of a converter according to claim 1, characterized in that, The parameterized constant admittance model is chosen as the mathematical model for simulation calculations of power electronic switching elements in power systems. Its basic formula is as follows: In the formula u (t), i (t) represents the voltage and current at the moment of branch solution; u (t-Δt), i (t-Δt) represents the voltage and current of the branch at the previous moment; I his (t) represents the equivalent historical current source; α , β These are the voltage coefficient and current coefficient of the equivalent current source, respectively; G eq This represents the equivalent conductivity matrix of the system.

3. The parallel optimization method for electromagnetic transient simulation of a converter according to claim 1, characterized in that, The parameterized constant admittance mathematical model for power electronic switching elements in the equivalent network adopts a cross-initialization algorithm to reduce transient errors during switching state transitions.

4. The parallel optimization method for electromagnetic transient simulation of a converter according to claim 3, characterized in that, The method to reduce transient errors during switch state switching is as follows: When there are mutually exclusive pairs of switch groups in the converter model, a cross-initialization algorithm is adopted. At the instant of switch state change, the electrical state quantity of the switch group on the other side is reinitialized with the previous step length of the electrical state quantity of the switch group. This serves as the basis for calculating the historical current source of the switch on this side. This further reduces the transient errors caused by harmonics and virtual power losses in the constant admittance model, and obtains simulation results that are closer to the ideal switch.

5. The parallel optimization method for electromagnetic transient simulation of a converter according to claim 1, characterized in that, The process of electromagnetic transient simulation calculation based on node analysis in a network representing a high proportion of power electronic devices in matrix form includes the following steps: Calculate the equivalent current source for each branch. （1） The superscript " n +1 and " n "Represents the quantities to be solved in the next moment and the quantities known in the previous moment, respectively." I his , U b , I b Let each branch be represented by a vector consisting of the equivalent current source, the voltage across the branch, and the current flowing through the branch, with a magnitude equal to the total number of branches. N b ; α , β For scale N b N b The diagonal matrix, if the first... i For each branch of the switching element in the strain gauge model, the diagonal matrix... α , β The first in i diagonal elements α ( i ), β ( i The value of ) is determined by the switch state and the following formula; In the formula α on , β on and α off , β off They respectively represent if the first i a side road α , β The value of the corresponding element in the matrix when the switch is on or off, if the first element... i Each branch corresponds to passive components such as inductors and capacitors. α ( i ), β ( i The value of ) is obtained using the nodal analysis method; G b The diagonal elements are the admittances of each branch, and the size is a diagonal matrix. N b N b , C 1. C 2 is the scale of N b N b The cross-initialization matrix is ​​used when all switch states in the next time step are the same as those in the previous time step. C 1. C Both are diagonal matrices with diagonal elements equal to 1. If in the next time step, the first... i The switch state corresponding to the branch changes, assuming it is related to the first branch. i The switches on the branch circuit form another switch in the same bridge arm located on the first branch circuit. j One branch, according to the definition of cross initialization: Calculate the injection current at the unknown node （2） In the formula I inj,u Let the vector be the injected current of all nodes with unknown voltages, and its size be the number of unknown nodes. N u , I s The current vector injected into the node by the external power source has a magnitude of N b ; M u The system correlation matrix for unknown nodes has a size of . N u N b For any branch i With nodes j , M u The value is defined as: Solving for the voltage at the unknown node （3） In the formula U n,u Let V be a vector of voltages at unknown nodes, with a magnitude of ... N u ; V n,k Let the vector be composed of the voltages of the known nodes, and its size be the number of known nodes. N k ; G uu and G uk The size of the unknown nodes is respectively N u N u The self-admittance matrix and the size of the unknown nodes compared to the known nodes are: N u N k The mutual admittance matrices are all part of the system admittance matrix; Solve for the voltage and current of each branch. （4） （5） In the formula U n,k To expand the size to N b The voltage vector of a known node.

6. The parallel optimization method for electromagnetic transient simulation of a converter according to claim 5, characterized in that, The values ​​of the diagonal elements of the diagonal matrix depend on the branch type and system parameters.

7. A device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement a parallel optimization method for converter electromagnetic transient simulation as described in any one of claims 1-6.

8. A storage medium containing computer-executable instructions, characterized in that, The computer-executable instructions, when executed by a computer processor, are used to perform a parallel optimization method for converter electromagnetic transient simulation as described in any one of claims 1-6.

Images