Parameter optimization method of metal oxide arrester based on W-MOPSO algorithm
The dynamic model parameters of metal oxide lightning arrester are optimized through the W-MOPSO algorithm, which solves the problem of insufficient pulse current adaptability in different wavefront times of the existing model, and improves the transient simulation accuracy and lightning arrester performance of the power system.
Patent Information
- Application Number
- CN202411031241.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-29
AI Technical Summary
The existing dynamic model of metal oxide lightning arrester has a small adaptation range when considering pulse currents at different wavefront times at the same time, and cannot truly reflect the actual situation, resulting in insufficient accuracy of transient simulation of the power system.
The parameter optimization method based on W-MOPSO algorithm is adopted to construct a single-pole pulse current generator model and a metal oxide lightning arrester dynamic model. By comprehensively considering the errors of steep slope impact, lightning impact and operating impact current residual voltage test, the model parameters are optimized, and the model parameters are optimized, so as to Reduce residual pressure test errors and use MATLAB and Simulink for joint simulation.
The more accurate simulation of pulse currents of different wavefront times by the dynamic model of metal oxide lightning arrester is realized, and the accuracy of transient simulation of the power system and the electrical performance of the lightning arrester are improved.
Smart Images

Figure CN119047299A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of power system simulation and design, and in particular to a parameter optimization method for a metal oxide surge arrester based on a W-MOPSO algorithm. Background Art
[0002] In power systems, common causes of power outages are tripping caused by lightning and switching overvoltage. Metal oxide arresters (MOAs) are currently widely used as protection devices against lightning and switching overvoltage. For the study of slow wave fronts, MOAs can be modeled through their nonlinear volt-ampere characteristics. [1,2] However, such a model is not suitable for fast front transient and lightning surge studies, because the dynamic characteristics obtained from the MOA show that the voltage across the arrester increases as the time for the arrester current to reach its peak decreases, and the arrester voltage reaches its peak before the arrester current reaches its maximum value. [1-3] Accurate modeling and simulation of arrester dynamic characteristics are very important for arrester configuration, system reliability assessment, and insulation coordination studies.
[0003] The dynamic characteristics obtained from the MOA show that the voltage across the arrester increases as the time for the arrester current to reach its peak decreases, and the arrester voltage reaches its peak before the arrester current reaches its maximum value. Generally, the residual voltage of a steep slope impulse current with a wavefront time of 1μs is 8%-12% higher than the residual voltage of a lightning impulse current with a wavefront time of 8μs. The residual voltage corresponding to the operating switch impulse with a longer time to reach the peak between 45 and 60μs is 2%-4% lower than the residual voltage of the lightning impulse current. [4-6] For this kind of arrester dynamic behavior, it is necessary to study a more complex arrester dynamic nonlinear model.
[0004] For the dynamic modeling of lightning arresters, the topologies used at home and abroad are mainly IEEE, Pinceti and Fernandez models. The IEEE model performs better in terms of modeling accuracy due to its high model complexity. In recent years, different procedures have been proposed to estimate the parameters of the model.
[0005] In reference [6], the authors proposed a numerical method to estimate the parameters of the three mentioned models; in references [7,8], the authors proposed a new procedure using a heuristic algorithm to determine the optimal parameters of the MOA model. The objective functions of these procedures all use the relative error between the simulation results and the experimental measurement results of the 8 / 20μs lightning impulse current residual voltage. However, they require the measurement results of the residual voltage under the 8 / 20μs pulse, and these actual measurement results are not always available. The complete test current and measured residual voltage data cannot be provided in the data sheet provided by the manufacturer. In order to overcome this problem, many literatures only consider the simulated residual voltage peak value that is more important for power system transient simulation. In reference [9], the authors proposed a procedure whose objective function is to minimize the relative error between the simulated residual voltage of the 8 / 20μs pulse current and the tested residual voltage.
[0006] However, in the standard for AC gapless metal oxide arresters, the residual voltage test needs to consider three tests, namely, the 1 / 10μs steep slope impact residual voltage test, the 8 / 20μs lightning impulse residual voltage test, and the 30 / 60μs operating impulse residual voltage test. Currently, there is no relevant literature that can consider the three residual voltage tests simultaneously when establishing the dynamic model of the arrester. This will cause the established MOA dynamic model to have a large difference in the effect of dealing with pulse currents with different wavefront times, resulting in a small range of model adaptability. Moreover, when studying pulse current problems in reality, it is often not a simple unipolar pulse current, but a complex multi-level pulse current with different wavefront times. If the MOA dynamic model can adapt to pulse currents with multiple wavefront times at the same time, it will be able to more realistically use the simulation results to reflect the actual situation and improve the accuracy of transient simulation of the power system.
[0007] References
[0008] [1]IEEE Working Group 3.4.11.Modeling of metal oxide surgearresters.IEEE Transactions on Power Delivery 1992;7(1):302e9.
[0009] [2]Pinceti P, Giannettoni MA simplified model for zinc oxide surgearresters. IEEE Transactions on Power Delivery 1999;14(2):393e8.
[0010] [3]Fernandez F,Diaz R.Metal oxide surge arrester model for fasttransient simulations.In:The Int.Conf.Onpower system transients IPAT’01,RioDe Janeiro,Brazil,2001.
[0011] [4]Martinez JA,Durbak DW.Parameter determination for modeling systemstransients-part V:surge arrester.IEEE Transactions on Power Delivery 2005;20(3):2073e8.
[0012] [5]Zinik B,Babuder M,Muhr M,Zitnik M,Thottappillil R.Numericalmodeling of metal oxide varistor.In:Proceedings of the XIVth Internationalsymposium on high voltage engineering.Beijing,China:Tsinghua University;August 25e29,2005.
[0013] [6]Li HJ,Birlasekaran S,Choi SS.A parameter identification techniquefor metaloxide surge arrester models.IEEE Transactions on Power Delivery2002;17(3):736e41.
[0014] [7]Bayadi A.Parameter identification of ZnO surge arrester modelsbased on genetic algorithms.Electric Power System Research 2008;78(7):1204e9.
[0015] [8]Nafar M, Gharehpetian GB, Niknam TA New parameter estimation algorithm for metal oxide surge arrester. Electric Power Components andSystems 2011;39(7):696e712.
[0016] [9]Christodoulou CA, Vita V, Ekonomou L, Chatzarakis GE, StathopulosIA. Application of Powell's optimization method to surge arrester circuitmodels parameters. Energy 2010; 35(8):3375e80. Summary of the invention
[0017] In view of at least one deficiency in the prior art, the present invention provides a parameter optimization method for a metal oxide arrester based on a W-MOPSO algorithm, which can optimize the parameters of a dynamic model of a metal oxide arrester and reduce the error of a residual voltage test. Then, the optimized parameters can be used to design the metal oxide arrester.
[0018] To achieve the above object, the present invention provides the following technical solutions:
[0019] A method for optimizing parameters of a metal oxide arrester comprises the following steps:
[0020] Constructing a unipolar pulse current generator model and a dynamic model of a metal oxide arrester with parameters to be optimized, wherein the unipolar pulse current generator model is configured to be able to adjust the wavefront time, and the unipolar pulse current generator model is used to apply a unipolar pulse current to the dynamic model of the metal oxide arrester;
[0021] Determine the parameters that need to be optimized for the dynamic model of the metal oxide surge arrester as decision variables of the W-MOPSO algorithm;
[0022] The residual voltage test error is used as the optimization objective function of the W-MOPSO algorithm, and the three test errors of steep slope impulse current residual voltage test, lightning impulse current residual voltage test, and switching impulse current residual voltage test are comprehensively considered as multi-objective optimization.
[0023] The W-MOPSO algorithm is used to optimize the dynamic model of the metal oxide surge arrester, and MATLAB and Simulink are used to perform joint simulation during the optimization process to obtain a parameter combination group of the optimized dynamic model of the metal oxide surge arrester;
[0024] The design parameters of the metal oxide arrester are determined according to the parameter combination group of the dynamic model of the metal oxide arrester.
[0025] The parameter optimization method of the metal oxide arrester as described above, further, the unipolar pulse current generator model specifically includes:
[0026] In Simulink, a pulse voltage generator is built to generate pulse voltage waveforms of different required waveforms;
[0027] Then the pulse voltage waveform is converted into a pulse current waveform through a voltage-controlled current source;
[0028] The pulse voltage generator includes a voltage source, an ignition gap, and a series resistor R f and load capacitor C L , charging capacitor C s And the parallel resistor R t , voltage source, ignition gap, series resistance R f and load capacitor C L Connect the charging capacitors C in series and then in parallel s And the parallel resistor R t .
[0029] In the parameter optimization method for the metal oxide arrester as described above, further, the pulse current waveform is given by the following preset formula:
[0030]
[0031]
[0032] I(t)=kU(t)
[0033] Among them, U0 is the DC high voltage source, α is the wavefront time correlation index, β is the wavetail time correlation index, and k is the proportional parameter.
[0034] The parameter optimization method of the metal oxide arrester as described above, further, the dynamic model of the metal oxide arrester specifically includes the IEEE arrester dynamic model, the Pinceti-Gianettoni arrester dynamic model and the Fernandez-Diaz arrester dynamic model:
[0035] The IEEE arrester dynamic model includes an inductor L0 and a resistor R0 connected in parallel, an inductor L1 and a resistor R1, a nonlinear resistor A0, and a nonlinear resistor A1 and a capacitor C connected in parallel, wherein the inductor L0 and the resistor R0 connected in parallel, the inductor L1 and the resistor R1 connected in parallel, and the nonlinear resistor A1 connected in series in sequence, and then the nonlinear resistor A0 and the capacitor C are connected in parallel;
[0036] The Pinceti-Gianettoni arrester dynamic model includes an inductor L0, an inductor L1, a nonlinear resistor A1, a nonlinear resistor A0 and a resistor R0, wherein the inductor L0, the inductor L1 and the nonlinear resistor A1 are sequentially connected in series, and then the nonlinear resistor A0 and the resistor R0 are connected in parallel;
[0037] The Fernandez–Diaz arrester dynamic model includes an inductor L0, a nonlinear resistor A1, a nonlinear resistor A0, a resistor R0 and a capacitor C, wherein the inductor L0 and the nonlinear resistor A1 are sequentially connected in series, and then the nonlinear resistor A0, the resistor R0 and the capacitor C are connected in parallel.
[0038] The parameter optimization method for the metal oxide surge arrester as described above, further, in the process of optimizing the dynamic model of the metal oxide surge arrester using the W-MOPSO algorithm, it is necessary to initialize the dynamic model of the metal oxide surge arrester, specifically including:
[0039] The calculation formula of the initial value of the parameters of the IEEE arrester dynamic model is as follows:
[0040] L0=0.2(d / n)μH
[0041] L1=15(d / n)μH
[0042] R0=100(d / n)Ω
[0043] R2=65(d / n)Ω
[0044] C=100(n / d)pF
[0045] Where n is the number of metal oxide disk rows and d is the overall height of the arrester;
[0046] The calculation formula of the initial value of the parameters of the Pinceti-Gianettoni arrester dynamic model is as follows:
[0047] R0=1MΩ
[0048] L0=(1 / 4)(U r )[(U r1 / T2 -U r8 / 20 ) / (U r8 / 20 )]
[0049] L1=(1 / 12)(U r )(U r1 / T2 -U r8 / 20 ) / (U r8 / 20 )]
[0050] Among them, U r1 / T2 is the residual voltage corresponding to the steep slope impulse current, U r8 / 20 It is the residual voltage corresponding to the lightning impulse current;
[0051] The calculation formula of the initial value of the parameters of the Fernandez–Diaz arrester dynamic model is as follows:
[0052] L0=nL'0
[0053] C = (100 / d)
[0054] R0=1MΩ
[0055] Where n is the scaling factor and L'0 is obtained by looking up the graph.
[0056] The parameter optimization method for the metal oxide surge arrester as described above further determines the parameters required to be optimized for the dynamic model of the metal oxide surge arrester as decision variables of the W-MOPSO algorithm, specifically,
[0057] The decision variables of the IEEE arrester dynamic model are [R0, L0, R1, L1, C], the decision variables of the Pinceti-Gianettoni arrester dynamic model are [R0, L0, L1], and the decision variables of the Fernandez–Diaz arrester dynamic model are
[0058] [R0, L0, C].
[0059] The parameter optimization method of the metal oxide arrester as described above, further, the residual voltage test error is used as the optimization objective function of the W-MOPSO algorithm, and the three test errors of the steep slope impulse current residual voltage test, the lightning impulse current residual voltage test, and the operation impulse current residual voltage test are comprehensively considered as multi-objective optimization, specifically,
[0060] Decision variables X = [X1, X2, X3, ...] T = [R ..., L ..., C] T ∈Ω, Ω is the feasible region of the decision variable;
[0061] f i (X) = abs(V sim-i (X,t)-V ref-i (t)) / V ref-i (t),i=1,2,3.
[0062] Where f1(X) represents the steep slope current residual voltage test error, f2(X) represents the lightning current residual voltage test error, f3(X) represents the switching impulse current residual voltage test error, V sim-i (X,t) represents the simulation results corresponding to the three tests, V ref-i (t) represents the measured results corresponding to the three tests;
[0063] The optimization problem is then expressed as follows:
[0064]
[0065] Where F(X) is the matrix composed of three test errors, F sum (X) is the weighted sum of the three test errors, ω i The linear weight coefficient is dynamically adjusted according to the optimization result. Each decision parameter Xi is set within the change range [X i-lower ,X i-upper ] is adjusted to minimize both F(X) and F sum (X), when the absolute value of each detection error is less than 0.04, the iterative calculation stops.
[0066] The parameter optimization method of the metal oxide surge arrester as described above further optimizes the dynamic model of the metal oxide surge arrester using the W-MOPSO algorithm, and performs joint simulation using MATLAB and Simulink during the optimization process, specifically,
[0067] Define the initial values of the parameters of the metal oxide arrester dynamic model in MATLAB, start the i-th iteration optimization, input the result of the i-th iteration optimization into Simulink for simulation calculation, obtain the input value of the fitness function in the i-th iteration process in Simulink, return the input value of the fitness function to MATLAB for fitness calculation, and then perform the evolution operation of the particle swarm to execute the i+1-th iteration.
[0068] Determine whether the i+1th iteration meets the preset number of iterations. If so, output the result to obtain the optimized parameter combination group of the metal oxide arrester dynamic model. If not, re-execute the i-th iteration optimization.
[0069] The parameter optimization method of the metal oxide surge arrester as described above, further, the W-MOPSO algorithm, specifically,
[0070] Initialize particle swarm and weight coefficients: randomly generate N particles in the decision variable space, each particle represents a combination of decision variables, and initialize the particle speed and individual optimal position; given the weight coefficient variation range, randomly initialize the weight coefficient combination;
[0071] Calculate the objective function value of the particle: calculate the value of each particle on all objective functions and save the value in the objective function value matrix;
[0072] Perform Pareto advantage sorting: Perform Pareto advantage sorting on all particles in the objective function value matrix, and then divide the particles into different advantage levels according to the results of Pareto advantage sorting;
[0073] According to the dominance level and weight coefficient of the particle, the individual best position and the global best position of the particle are updated: an individual best position is selected according to the dominance level of each particle, and then the individual best position and the global best position are weighted averaged according to the weight coefficient to obtain a new best position;
[0074] Update particle speed and position: Update the speed of each particle based on its current position and individual best position;
[0075] Check whether the termination condition is met: If the termination condition is met, stop the algorithm and output the non-dominated solution set; otherwise, re-execute the step of calculating the objective function value of the particle.
[0076] Compared with the prior art, the present invention has the following beneficial effects:
[0077] 1. The present invention can simultaneously consider three residual voltage tests when establishing a dynamic model of a lightning arrester, can more realistically utilize simulation results to reflect actual conditions, and improve the accuracy of transient simulation of power systems.
[0078] 2. The present invention comprehensively considers multiple optimization objectives and multiple different dynamic models of lightning arresters, and improves the accuracy of transient simulation calculation of the dynamic model of the metal oxide lightning arrester in responding to pulse currents with different wavefront times.
[0079] 3. The present invention can perform Pareto dominance sorting on the global optimal solution of the weighted sum of multi-objective functions through the W-MOPSO algorithm, wherein the weighted sum is dynamically adjusted to solve the problem of optimization target bias in the multi-objective optimization process, and obtain a Pareto optimal solution set.
[0080] 4. The present invention utilizes the optimized parameters of the dynamic model of the metal oxide arrester to design the metal oxide arrester, which can improve the electrical performance of the metal oxide arrester. BRIEF DESCRIPTION OF THE DRAWINGS
[0081] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0082] Figure 1 is a flow chart of a method in an embodiment of the present invention;
[0083] Figure 2 is a circuit schematic diagram of a single-stage pulse current generator in an embodiment of the present invention;
[0084] Figure 3 is a circuit schematic diagram of the IEEE arrester dynamic model in an embodiment of the present invention;
[0085] Figure 4 is a circuit schematic diagram of a dynamic model of a Pinceti-Gianettoni arrester in an embodiment of the present invention;
[0086] Figure 5 is a circuit schematic diagram of a dynamic model of a Fernandez–Diaz arrester in an embodiment of the present invention;
[0087] Figure 6 Schematic diagram of typical VI characteristics of metal oxide valve plates of lightning arresters in an embodiment of the present invention;
[0088] Figure 7 It is a flow chart of MATLAB and Simulink joint simulation optimization parameters in an embodiment of the present invention;
[0089] Figure 8 It is a diagram showing the change of the objective function value in the algorithm iteration process in the embodiment of the present invention;
[0090] Fig. 9 This is a simulation result diagram of a steep slope current residual voltage test in an embodiment of the present invention;
[0091] Fig.10 This is a simulation result diagram of a lightning current residual voltage test in an embodiment of the present invention;
[0092] Fig.11 The simulation result of the operating current residual voltage test in the embodiment of the present invention;
[0093] Fig.12 It is a diagram of the fitness values of different algorithm optimization parameters corresponding to multiple objectives in the embodiment of the present invention. DETAILED DESCRIPTION
[0094] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0095] Example:
[0096] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations of the embodiments of the present invention are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0097] In the description of the present invention, "multiple" means at least two, such as two, three, etc., unless otherwise clearly defined. In addition, unless otherwise clearly defined and defined, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0098] In order to accurately estimate the parameters of the arrester model and make the arrester dynamic model applicable to pulse currents with different wavefront times, including lightning impulse simulation, steep slope impulse simulation and operating switch impulse transient simulation, this invention proposes a multi-objective particle swarm optimization based on weighted Pareto-dominance sort (W-MOPSO) optimization method, which can optimize the three simulation errors at the same time and obtain a Pareto optimal solution set. The model parameters obtained by this solution set can make the arrester dynamic model applicable to pulse currents with different wavefront times at the same time. Finally, MATLAB / Simulink was used to simulate the residual voltage test of the dynamic model of MOA, and the simulation results were applied to the developed W-MOPSO algorithm program to determine the parameters of different models. The validity and accuracy of the estimated parameters were evaluated by comparing the predicted residual voltage with the experimental results. The dynamic model of the arrester optimized by the proposed algorithm can make the absolute values of the relative average errors of multiple optimization objectives less than 4%. Using the optimized parameters to design the dynamic model of the metal oxide arrester can simultaneously improve the accuracy of the residual pressure simulation results of the metal oxide arrester when dealing with slow and fast pulse currents, thereby effectively promoting the optimization of the dynamic arrester modeling parameters and ensuring the accuracy of the model.
[0099] See also Figure 1 , Figure 1 The method flow chart in the embodiment of the present invention is shown. A parameter optimization method for a metal oxide surge arrester provided in the embodiment of the present invention may include the following steps:
[0100] Step 100: construct a unipolar pulse current generator model and a metal oxide arrester dynamic model of parameters to be optimized, wherein the unipolar pulse current generator model is configured to be able to adjust the wavefront time, and the unipolar pulse current generator model is used to apply a unipolar pulse current to the metal oxide arrester dynamic model.
[0101] In this step, a unipolar pulse current generator model capable of adjusting the wavefront time is established. Specifically, the pulse current waveform generator can be designed based on the pulse current waveform calculation formula.
[0102] Specifically, in Simulink, a pulse voltage generator can be built to generate pulse voltage waveforms of different required waveforms. The voltage waveform is then converted into a current waveform through a voltage-controlled current source to obtain different pulse current waveforms. Among them, the pulse voltage generator uses high-voltage direct current as a voltage source, and capacitors and resistors of different values are required. A spherical gap is used to isolate the electrodes, and the two electrodes are short-circuited through sparks, discharging in a very short time.
[0103] Figure 2 The circuit schematic diagram of the unipolar pulse current generator in the embodiment of the present invention is shown. Figure 2 A unipolar pulse current can be obtained in the generating circuit shown, and the current waveform parameters are defined by the following formula.
[0104]
[0105] I(t)=kU(t)
[0106] Among them, U0 is a DC high voltage source. Capacitor C s It is called charging capacitor, C L Called the load capacitor, the charging capacitor C s For charging from the DC supply voltage, the load capacitor C L For discharge. There are two waveform shaping resistors, the series resistor R f is the wavefront related resistance, the parallel resistance R t is the wave tail related resistance. α is the wave front time related index, and β is the wave tail time related index. By controlling the current source with voltage, changing the proportional parameter k, the voltage U(t) is converted into the desired I(t). By optimizing the model parameters of the single-stage pulse current generator, 1 / 10μs steep slope impulse, 8 / 20μs lightning impulse, and 30 / 60μs operating impulse current can be obtained.
[0107] In the process of establishing the dynamic model of the metal oxide arrester with parameters to be optimized, the corresponding model diagram and modeling process are as follows:
[0108] See also Figure 3 , Figure 3 The circuit schematic diagram of the IEEE lightning arrester dynamic model in the embodiment of the present invention is shown. It can be seen from the figure that the IEEE lightning arrester dynamic model includes an inductor L0 and a resistor R0 connected in parallel, an inductor L1 and a resistor R1, a nonlinear resistor A0 and a nonlinear resistor A1 and a capacitor C connected in parallel, wherein the inductor L0 and the resistor R0 connected in parallel, the inductor L1 and the resistor R1 connected in parallel, and the nonlinear resistor A1 are connected in series in sequence, and then the nonlinear resistor A0 and the capacitor C are connected in parallel.
[0109] See also Figure 4 , Figure 4The circuit schematic diagram of the Pinceti-Gianettoni arrester dynamic model in the embodiment of the present invention is shown. As can be seen from the figure, the Pinceti-Gianettoni arrester dynamic model includes an inductor L0, an inductor L1, a nonlinear resistor A1, a nonlinear resistor A0 and a resistor R0, wherein the inductor L0, the inductor L1 and the nonlinear resistor A1 are sequentially connected in series, and then the nonlinear resistor A0 and the resistor R0 are connected in parallel;
[0110] See also Figure 5 , Figure 5 The circuit schematic diagram of the Fernandez–Diaz arrester dynamic model in the embodiment of the present invention is shown. It can be seen from the figure that the Fernandez–Diaz arrester dynamic model includes an inductor L0, a nonlinear resistor A1, a nonlinear resistor A0, a resistor R0 and a capacitor C, wherein the inductor L0 and the nonlinear resistor A1 are connected in series in sequence, and then the nonlinear resistor A0, the resistor R0 and the capacitor C are connected in parallel.
[0111] The above three models all contain two nonlinear resistors and different numbers of resistors, inductors, and capacitors.
[0112] In the process of constructing the nonlinear resistor module, the nonlinear VI characteristic adjustment of the nonlinear resistors A0 and A1 needs to be based on the reference values in Table 1.
[0113] Table 1 VI characteristics of A0 and A1
[0114]
[0115] The definition of nonlinear resistance in Simulink is as follows. The nonlinear VI characteristics of each column of the arrester are modeled by a combination of three exponential functions of the following form:
[0116]
[0117] Among them, V and I are the voltage across the arrester model and the current flowing through the arrester respectively. ref is the protection voltage, which is the reference current I ref The nonlinear VI characteristic curve mainly divides the current into three regions. The definition function needs to be set to three nonlinear curve functions. Typical VI characteristics are as follows: Figure 6 As shown. The nonlinear parameter k i and α i It needs to be adjusted according to the actual VI characteristics and does not change with the protection voltage. :
[0118] Nonlinear parameter k i and α iThe definition method is as follows, corresponding to the numbers 1, 2 and 3, representing the three current regions, and the letter n is the number of metal oxide disks in parallel. The conversion current i1 from region 1 to region 2 and the conversion current i2 from region 2 to region 3 are calculated as follows:
[0119]
[0120] Among them, the calculation of the conversion parameter Pi of the conversion current is related to the interval, and the calculation method is as follows:
[0121]
[0122] According to the above formula, combined with the VI characteristic data of the nonlinear resistors A0 and A1, the parameters k and α can be obtained by using the least squares method, so that the nonlinear resistor module can be constructed.
[0123] The calculation formulas for the initial values of the IEEE model R0, L0, R1, L1 and C parameters are shown below, taking into account the height and number of arrester columns. The letter n is the number of metal oxide disc columns and d is the overall height of the arrester.
[0124] L0=0.2(d / n)μH
[0125] L1=15(d / n)μH
[0126] R0=100(d / n)Ω
[0127] R2=65(d / n)Ω
[0128] C=100(n / d)pF
[0129] The calculation formula for the initial values of the Pinceti-Gianettoni model parameters is as follows.
[0130] R0=1MΩ
[0131] L0=(1 / 4)(U r )[(U r1 / T2 -U r8 / 20 ) / (U r8 / 20 )]
[0132] L1=(1 / 12)(U r )(U r1 / T2 -U r8 / 20 ) / (U r8 / 20 )]
[0133] Among them, U r1 / T2 is the residual voltage corresponding to the steep slope impulse current, U r8 / 20 It is the residual voltage corresponding to the lightning impulse current.
[0134] The calculation formula for the initial values of the Fernandez–Diaz model parameters is as follows.
[0135] L0=nL'0
[0136] C = (100 / d)
[0137] R0=1MΩ
[0138] Where n is the scaling factor and L'0 is obtained by looking up the graph.
[0139] Step 200: Determine the parameters that need to be optimized for the metal oxide surge arrester dynamic model as decision variables of the W-MOPSO algorithm.
[0140] In this step, the parameters that need to be optimized in the dynamic model are determined as the decision variables of the W-MOPSO algorithm. The decision variables of the IEEE model are [R0, L0, R1, L1, C], the decision variables of the Pinceti-Gianettoni model are [R0, L0, L1], and the decision variables of the Fernandez–Diaz model are [R0, L0, C].
[0141] Step 300: Taking the residual voltage test error as the optimization objective function of the W-MOPSO algorithm, comprehensively considering the three test errors of the steep slope impulse current residual voltage test, the lightning impulse current residual voltage test, and the switching impulse current residual voltage test as multi-objective optimization.
[0142] In this step, the three test errors of steep slope impulse current residual voltage test, lightning impulse current residual voltage test and switching impulse current residual voltage test are comprehensively considered as multi-objectives for optimization;
[0143] Decision variables X = [X1, X2, X3, …] T =[R…,L…,C] T ∈Ω, the specific form of the decision variable is defined according to step 200, and Ω is the feasible region of the decision variable.
[0144] f i (X) = abs(V sim-i (X,t)-V ref-i (t)) / V ref-i (t),i=1,2,3.
[0145] Where f1(X) represents the steep slope current residual voltage test error, f2(X) represents the lightning current residual voltage test error, f3(X) represents the switching impulse current residual voltage test error, V sim-i (X,t) represents the simulation results corresponding to the three tests, V ref-i (t) represents the measured results corresponding to the three tests.
[0146] The optimization problem is then expressed as follows:
[0147]
[0148] F(X) is the matrix composed of three test errors, F sum (X) is the weighted sum of the three test errors, ω i is the linear weight coefficient dynamically adjusted according to the optimization results. Each decision parameter X i In the setting change interval [X i-lower ,X i-upper ] is adjusted to minimize both F(X) and F sum (X), when the absolute value of each detection error is less than 0.04, the iterative calculation stops.
[0149] Step 400: Optimize the dynamic model of the metal oxide surge arrester using the W-MOPSO algorithm, and perform joint simulation using MATLAB and Simulink during the optimization process to obtain a parameter combination group of the optimized dynamic model of the metal oxide surge arrester.
[0150] In this step, Figure 7 The flowchart of the MATLAB and Simulink joint simulation optimization parameters in the embodiment of the present invention is shown. Figure 7 It can be seen that the initial values of the parameters of the metal oxide arrester dynamic model are defined in MATLAB, the i-th iteration optimization is started, the result of the i-th iteration optimization is input into Simulink for simulation calculation, the input value of the fitness function in the i-th iteration process is obtained in Simulink, the input value of the fitness function is returned to MATLAB for fitness calculation, and then the particle swarm is evolved to execute the i+1-th iteration.
[0151] It is determined whether the i+1th iteration meets the preset number of iterations. If so, the result is output to obtain the optimized parameter combination group of the metal oxide arrester dynamic model. If not, the i-th iteration optimization is re-executed.
[0152] The specific process of the W-MOPSO algorithm is as follows:
[0153] 1) Initialize the particle swarm and weight coefficients. Randomly generate N particles in the decision variable space, each particle represents a feasible solution, that is, a combination of decision variables. At the same time, initialize the particle speed and individual optimal position. Specifically, for the i-th particle, its position can be expressed as Its speed can be expressed as Its individual optimal position can be expressed as Where n represents the dimension of the decision variable. Given the weight coefficient variation interval, the weight coefficient combination is randomly initialized.
[0154] 2) Calculate the objective function value of the particle. For each particle, calculate its value on all objective functions and save it in an objective function value matrix. Assuming there are M objective functions, the objective function value matrix can be expressed as F = (f ij ) N×M , where f ij Represents the value of the i-th particle on the j-th objective function.
[0155] 3) Perform Pareto dominance sorting. Perform Pareto dominance sorting on all particles in the objective function value matrix and divide them into different dominance levels. In Pareto dominance sorting, if a particle is better than another particle in all objective functions, it means that the particle dominates the other particle. All particles that are not dominated belong to the first dominance level, and then the particles dominated by the particles of the first dominance level are classified as the second dominance level, and so on, until all particles are divided into different dominance levels.
[0156] 4) Update the individual best position and global best position of the particle according to the dominance level and weight coefficient of the particle: For each particle, first select an individual best position according to its dominance level, then perform a weighted average of it and the global best position according to the weight coefficient to obtain a new best position. Specifically, for each particle i, its individual best position p i and the global optimal position g i The update method is as follows:
[0157] If particle i is at the highest advantage level, its current position is set to the individual best position, i.e., p i =x i .
[0158] Otherwise, select an individual best position from a particle with a higher advantage level than particle i as the individual best position of particle i, that is, p i =p i , where i′ is a randomly selected particle with a higher advantage level than particle i.
[0159] The individual best positions of all particles are weighted averaged according to the weight coefficient of each objective function to obtain the global best position, that is, Where m is the number of objective functions, w j is the weight coefficient of the jth objective function.
[0160] In the W-MOPSO algorithm, the weight coefficient in the objective function is a dynamic parameter, which will be adjusted continuously during the algorithm iteration process to better balance the contradictions between different objectives. Specifically, in each iteration process, the W-MOPSO algorithm will first perform non-dominated sorting on the particles according to the current particle swarm and the value of the objective function. In each iteration process, the global best position refers to the set of Pareto optimal solutions among all particles. The global best position needs to be continuously updated and the Pareto optimal solutions therein need to be saved in an external archive. Then, the algorithm will calculate the weight coefficient of each objective function based on the Pareto optimal solution in the external archive and the current value of the objective function. The calculation of the weight coefficient is usually based on the distance indicator, using the average value of all the best solutions found in history instead of the ideal solution, accumulating the Euclidean distance between each solution in the Pareto optimal solution set and an ideal solution (that is, the solution with the smallest value of all objective functions), and then using the accumulated sum as the weight coefficient.
[0161] 5) Update the speed and position of the particle. For each particle, first update its speed based on its current position and individual best position. Specifically, for the i-th particle, its speed update formula is:
[0162]
[0163] p ij (t+1)=p ij (t)+v ij (t+1)
[0164] Among them, wv ij (t) represents the inertia weight, which is the vector that continues the particle's previous motion; represents particle cognition, indicating that a single particle moves toward the optimal position; It represents group cognition and the particle swarm moving toward the global optimal position.
[0165] 6) Check whether the termination condition is met: If the termination condition is met, such as the maximum number of iterations or each objective function reaches the optimization requirement, the algorithm is stopped and the non-dominated solution set is output. Otherwise, return to step 2) and continue iterative optimization.
[0166] Step 500: Determine the design parameters of the metal oxide surge arrester according to the parameter combination group of the dynamic model of the metal oxide surge arrester.
[0167] In this step, the metal oxide arrester is designed using the parameters of the optimized metal oxide arrester dynamic model, which can improve the electrical performance of the metal oxide arrester.
[0168] The beneficial effects of the technical solution of the embodiment of the present invention will be described below in conjunction with simulation results.
[0169] The present invention proposes a parameter optimization method for metal oxide lightning arresters based on the W-MOPSO algorithm. The W-MOPSO algorithm can perform Pareto dominance sorting on the global optimal solution of the weighted sum of multi-objective functions, wherein the weighted sum is dynamically adjusted to solve the problem of partiality of optimization targets in the multi-objective optimization process, and a Pareto optimal solution set is obtained. Finally, the model of the YH5WZ-17 / 45 model lightning arrester is taken as an example to carry out model construction and parameter optimization analysis, and the optimization effects of various different algorithms are compared by comparing the predicted residual pressure with the experimental results. The dynamic model of the lightning arrester optimized by the proposed algorithm can make the absolute values of the relative average errors of multiple optimization targets less than 4%, which effectively promotes the optimization of the dynamic lightning arrester modeling parameters and ensures the accuracy of the model.
[0170] Taking the IEEE arrester dynamic model as an example, simulation analysis is performed. In the iterative process, the values of the three objective functions corresponding to the Pareto solution set change as follows: Figure 8 As shown in the figure, at the beginning, the values of the three objective functions are highly dispersed, and some of the values of individual objective functions are particularly large. As the iterations proceed, the values of each objective function become smaller and tend to be stable. At this time, the solution obtained is the Pareto solution set.
[0171] The simulation results of residual pressure test are as follows: Figures 9 to 11 , find one of the feasible solutions in the Pareto solution set, and get the optimized parameters and relative errors of the arrester, as shown in Table 2. The errors of the three tests are all less than 4%, and the optimization effect is good.
[0172] Table 2 Simulation results of residual pressure test after optimizing parameters
[0173]
[0174]
[0175] Finally, simulations are performed based on NSGA-II, MOEA / D, MOGWO and W-MOPSO. The effectiveness and superiority of the proposed W-MOPSO in solving the dynamic model parameter optimization of the arrester are evaluated based on the simulation results. The fitness values obtained by different algorithms are shown in Figure 2. Fig.12 As shown in the figure, based on NSGA-II, MOEA / D, and MOGWO algorithms, some fitness values are too large, and the dispersion is large, and the optimization effect is average. However, W-MOPSO can overcome the above shortcomings, and the obtained Pareto solution set achieves the best effect in almost all optimization objectives, which means that W-MOPSO is better than other multi-objective optimization methods.
[0176] The simulation results are summarized in Table 3, which is shown below:
[0177] Table 3 Optimization results obtained by different algorithms
[0178]
[0179] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.
[0180] The above embodiments are only for illustrating the technical concept and features of the present invention, and their purpose is to enable ordinary technicians in the field to understand the content of the present invention and implement it accordingly, and they cannot be used to limit the protection scope of the present invention. Any equivalent changes or modifications made based on the essence of the content of the present invention should be included in the protection scope of the present invention.
Claims
1. A parameter optimization method for a metal oxide surge arrester, characterized in that: include: Constructing a unipolar pulse current generator model and a dynamic model of a metal oxide arrester with parameters to be optimized, wherein the unipolar pulse current generator model is configured to be able to adjust the wave front and wave tail time, and the unipolar pulse current generator model is used to apply a unipolar pulse current to the dynamic model of the metal oxide arrester; Determine the parameters that need to be optimized for the dynamic model of the metal oxide surge arrester as decision variables of the W-MOPSO algorithm; The residual voltage test error is used as the optimization objective function of the W-MOPSO algorithm, and the three test errors of steep slope impulse current residual voltage test, lightning impulse current residual voltage test and switching impulse current residual voltage test are comprehensively considered as multi-objective optimization; The W-MOPSO algorithm is used to optimize the dynamic model of the metal oxide surge arrester, and MATLAB and Simulink are used to perform joint simulation during the optimization process to obtain a parameter combination group of the optimized dynamic model of the metal oxide surge arrester; The design parameters of the metal oxide arrester are determined according to the parameter combination group of the dynamic model of the metal oxide arrester.
2. The parameter optimization method of a metal oxide surge arrester according to claim 1, characterized in that: The unipolar pulse current generator model specifically includes: In Simulink, a pulse voltage generator is built to generate pulse voltage waveforms of different required waveforms; Then the pulse voltage waveform is converted into a pulse current waveform through a voltage-controlled current source; The pulse voltage generator includes a voltage source, an ignition gap, and a series resistor R f and load capacitor C L , charging capacitor C s And the parallel resistor R t , voltage source, ignition gap, series resistance R f and load capacitor C L Connect the charging capacitors C in series and then in parallel s And the parallel resistor R t .
3. The parameter optimization method of a metal oxide surge arrester according to claim 2, characterized in that: The pulse current waveform is determined by the following preset formula: I(t)=kU(t) Among them, U0 is the DC high voltage source, α is the wavefront time correlation index, β is the wavetail time correlation index, and k is the proportional parameter.
4. The parameter optimization method for metal oxide surge arrester according to claim 1, characterized in that: The metal oxide arrester dynamic model specifically includes the IEEE arrester dynamic model, the Pinceti-Gianettoni arrester dynamic model and the Fernandez-Diaz arrester dynamic model: The IEEE arrester dynamic model includes an inductor L0 and a resistor R0 connected in parallel, an inductor L1 and a resistor R1, a nonlinear resistor A0, and a nonlinear resistor A1 and a capacitor C connected in parallel, wherein the inductor L0 and the resistor R0 connected in parallel, the inductor L1 and the resistor R1 connected in parallel, and the nonlinear resistor A1 connected in series in sequence, and then the nonlinear resistor A0 and the capacitor C are connected in parallel; The Pinceti-Gianettoni arrester dynamic model includes an inductor L0, an inductor L1, a nonlinear resistor A1, a nonlinear resistor A0 and a resistor R0, wherein the inductor L0, the inductor L1 and the nonlinear resistor A1 are sequentially connected in series, and then the nonlinear resistor A0 and the resistor R0 are connected in parallel; The Fernandez–Diaz arrester dynamic model includes an inductor L0, a nonlinear resistor A1, a nonlinear resistor A0, a resistor R0 and a capacitor C, wherein the inductor L0 and the nonlinear resistor A1 are sequentially connected in series, and then the nonlinear resistor A0, the resistor R0 and the capacitor C are connected in parallel.
5. The parameter optimization method for metal oxide surge arrester according to claim 4, characterized in that: In the process of optimizing the dynamic model of the metal oxide surge arrester by using the W-MOPSO algorithm, the dynamic model of the metal oxide surge arrester needs to be initialized, which specifically includes: The calculation formula of the initial value of the parameters of the IEEE arrester dynamic model is as follows: L0=0.2(d / n)μH L1=15(d / n)μH R0=100(d / n)Ω R2=65(d / n)Ω C=100(n / d)pF Where n is the number of metal oxide disk rows and d is the overall height of the arrester; The calculation formula of the initial value of the parameters of the Pinceti-Gianettoni arrester dynamic model is as follows: R0=1MΩ L0=(1 / 4)(U) r )[(IN r1 / T2 -IN r8 / 20 ) / (IN r8 / 20 )] L1=(1 / 12)(U) r )(IN r1 / T2 -IN r8 / 20 ) / (IN r8 / 20 )] Among them, U r1 / T2 is the residual voltage corresponding to the steep slope impulse current, U r8 / 20 It is the residual voltage corresponding to the lightning impulse current; The calculation formula of the initial value of the parameters of the Fernandez–Diaz arrester dynamic model is as follows: L0=nL′0 C = (100 / d) R0=1MΩ Where n is the scaling factor and L'0 is obtained by looking up the graph.
6. The parameter optimization method for a metal oxide surge arrester according to claim 5, characterized in that: Determine the parameters that need to be optimized for the dynamic model of the metal oxide surge arrester as decision variables of the W-MOPSO algorithm, specifically, The decision variables of the IEEE arrester dynamic model are [R0, L0, R1, L1, C], the decision variables of the Pinceti-Gianettoni arrester dynamic model are [R0, L0, L1], and the decision variables of the Fernandez–Diaz arrester dynamic model are [R0, L0, C].
7. The parameter optimization method for a metal oxide surge arrester according to claim 6, characterized in that: The residual voltage test error is used as the optimization objective function of the W-MOPSO algorithm, and the three test errors of steep slope impulse current residual voltage test, lightning impulse current residual voltage test, and switching impulse current residual voltage test are comprehensively considered as multi-objective optimization. Specifically, Decision variables X = [X1, X2, X3, …] T =[R…,L…,C] T ∈Ω, Ω is the feasible region of the decision variable; f i (X)=abs(V sim-i (X,t)-V ref-i (t)) / V ref-i (t),i=1,2,3. Where f1(X) represents the steep slope current residual voltage test error, f2(X) represents the lightning current residual voltage test error, f3(X) represents the switching impulse current residual voltage test error, V sim-i (X,t) represents the simulation results corresponding to the three tests, V ref-i (t) represents the measured results corresponding to the three tests; The optimization problem is then expressed as follows: Where F(X) is the matrix composed of three test errors, F sum (X) is the weighted sum of the three test errors, ω i The linear weight coefficient is dynamically adjusted according to the optimization result. Each decision parameter Xi is set within the change range [X i-lower ,X i-upper ] is adjusted to minimize both F(X) and F sum (X), when the absolute value of each detection error is less than 0.04, the iterative calculation stops.
8. The parameter optimization method for metal oxide surge arrester according to claim 7, characterized in that: The W-MOPSO algorithm is used to optimize the dynamic model of the metal oxide surge arrester. During the optimization process, MATLAB and Simulink are used for joint simulation. Specifically, Define the initial values of the parameters of the metal oxide arrester dynamic model in MATLAB, start the i-th iteration optimization, input the result of the i-th iteration optimization into Simulink for simulation calculation, obtain the input value of the fitness function in the i-th iteration process in Simulink, return the input value of the fitness function to MATLAB for fitness calculation, and then perform the evolution operation of the particle swarm to execute the i+1-th iteration. It is determined whether the i+1th iteration meets the preset number of iterations. If so, the result is output to obtain the optimized parameter combination group of the metal oxide arrester dynamic model. If not, the i-th iteration optimization is re-executed.
9. The parameter optimization method for a metal oxide surge arrester according to claim 8, characterized in that: The W-MOPSO algorithm is specifically, Initialize particle swarm and weight coefficients: randomly generate N particles in the decision variable space, each particle represents a combination of decision variables, and initialize the particle speed and individual optimal position; given the weight coefficient variation range, randomly initialize the weight coefficient combination; Calculate the objective function value of the particle: calculate the value of each particle on all objective functions and save the value in the objective function value matrix; Perform Pareto advantage sorting: Perform Pareto advantage sorting on all particles in the objective function value matrix, and then divide the particles into different advantage levels according to the results of Pareto advantage sorting; According to the dominance level and weight coefficient of the particle, the individual best position and the global best position of the particle are updated: an individual best position is selected according to the dominance level of each particle, and then the individual best position and the global best position are weighted averaged according to the weight coefficient to obtain a new best position; Update particle speed and position: Update the speed of each particle based on its current position and individual best position; Check whether the termination condition is met: If the termination condition is met, stop the algorithm and output the non-dominated solution set; otherwise, re-execute the step of calculating the objective function value of the particle.
Citation Information
Cited By
External multi-gap lightning arrester volt-second characteristic curve estimation method, device and medium
CN119757996A
Lightning arrester dynamic optimization algorithm and device based on improved genetic algorithm
CN120145712A
A Lightning Arrester Dynamic Optimization Algorithm and Device Based on Improved Genetic Algorithm
CN120145712B
