Parameter self-tuning method for short-circuit breaking test loop of high-voltage alternating-current circuit breaker

Abstract

The invention discloses a parameter self-tuning method for a short-circuit breaking test loop of a high-voltage alternating-current circuit breaker, which belongs to the technical field of high-voltage electrical equipment test of a power system, and is characterized in that a three-phase direct test loop is equivalent to a single-phase model, and voltage and current standard values and time angle domain conversion are introduced; establishing a digital twinning differential equation model under dimensionless parameters; secondly, taking dimensionless parameters as decision variables of an improved particle swarm optimization algorithm, dynamically adjusting an inertia weight through a state sensing mechanism to balance global exploration and local development capabilities, and giving a targeted flight direction to particles in combination with a multi-source error compensation strategy; in the design of a value function, a layered adaptive penalty mechanism is adopted to preferentially guarantee the satisfaction of hard constraints such as an amplitude coefficient and the like, and transient recovery voltage waveform characteristic parameters are automatically extracted through a coordinate conversion method for simulation verification. According to the invention, high-efficiency, accurate and automatic setting of high-voltage AC circuit breaker short-circuit breaking test loop parameters is finally realized.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of high-voltage electrical equipment test of power system, and particularly relates to a parameter self-tuning method for short-circuit breaking test loop of high-voltage AC circuit breaker. BACKGROUND

[0002] The high-voltage AC circuit breaker is the core protection equipment for ensuring the safe and stable operation of the power system, and its breaking performance must be verified by strict large-capacity tests. Both the international standard IEC 62271-100 and the Chinese national standard GB / T 1984-2014 have made clear requirements for short-circuit breaking tests, among which the direct test is widely adopted because it can best reflect the actual operating conditions. In this test, the configuration of the loop parameters directly determines the shape of the transient recovery voltage waveform, which must fully meet the envelope requirements specified in the standard, otherwise the actual breaking capacity of the circuit breaker cannot be effectively evaluated.

[0003] At present, the parameter tuning work of high-voltage test loop still faces significant challenges. The traditional method mainly relies on the accumulation of engineers' experience and repeated trial and error, by manually adjusting inductance, capacitance, resistance and other parameters, and performing multiple simulation verifications. This process not only consumes time and effort, and has a long cycle, but also cannot guarantee the optimal parameter combination, especially when facing multiple test projects such as T10, T30, T60, T100a, T100s, etc., the problems of low efficiency and insufficient precision are more prominent.

[0004] Although there have been attempts to introduce standard optimization algorithms to achieve automatic parameter tuning, such as using the standard particle swarm algorithm, there are still obvious limitations in practical application. High-voltage test loop has strong non-linear characteristics, which leads to the traditional optimization algorithm generally appearing premature convergence, blind search direction, excessive iteration times and other problems when dealing with such complex problems. More importantly, the standard algorithm lacks intelligent processing capability for engineering constraints, especially it is difficult to accurately meet the mandatory requirements such as "amplitude coefficient shall not be lower than the standard value", often leading to the optimization result being mathematically optimal but not engineering feasible.

[0005] Therefore, the existing technology cannot provide a parameter tuning scheme that can balance efficiency, precision and engineering applicability, which seriously restricts the automation process and reliability improvement of high-voltage electrical test. Developing a special tuning method with self-adaptive adjustment mechanism, which can intelligently balance exploration and development, and can effectively embed engineering constraints, has become a technical problem urgently to be solved in the field. SUMMARY

[0006] The present application provides a parameter self-setting method for a short-circuit breaking test loop of a high-voltage alternating current circuit breaker, which is used for parameter self-setting of a three-phase direct test loop of a high-voltage alternating current circuit breaker.

[0007] To solve the above technical problems, the present application adopts the following technical solutions:

[0008] A parameter self-setting method for a short-circuit breaking test loop of a high-voltage alternating current circuit breaker, which is used for parameter self-setting of a three-phase direct test loop of a high-voltage alternating current circuit breaker.

[0009] Step one: according to symmetry, the three-phase direct test loop is equivalent to a single-phase equivalent test loop, and the phase where the first breaking circuit breaker is located is taken as the equivalent single-phase, and the transformer and the reactor in the three-phase direct test loop are equivalent to an inductor, and the voltage source is an equivalent voltage source, and an equivalent model is established;

[0010] Step two: according to Kirchhoff's current law (KCL), Kirchhoff's voltage law (KVL) and the characteristics of passive elements, the time-domain equation set and initial conditions of the equivalent model are written;

[0011] Step three: the standard values of voltage and current are selected for dimensionless transformation, and the time domain is converted to the angle domain, and the equation set and initial conditions are rewritten to obtain the equivalent model under the dimensionless parameters;

[0012] Step four, according to the equation set obtained in step three, the differential equation relationship describing the voltage per unit of the test circuit breaker and other dimensionless parameters is obtained;

[0013] Step five, the dimensionless parameters obtained in step three are taken as the decision variables of the improved particle swarm optimization algorithm (IPSO), and the initial parameters of the improved particle swarm optimization algorithm are set;

[0014] Step six, according to the differential equation relationship obtained in step four, the value function of the improved particle swarm optimization algorithm is determined, and the value function adopts a hierarchical adaptive mechanism;

[0015] Step seven, the improved particle swarm optimization algorithm is iteratively run to find the dimensionless parameter combination that minimizes the value function, and the actual inductance, capacitance and resistance values of the test circuit are obtained through formula conversion.

[0016] Further, the time domain equation set and initial conditions of the equivalent model written in step two are:

[0017] ,

[0018] In the formula, represents time, and are the currents flowing through the equivalent inductance and the frequency modulation capacitance , respectively, is the voltage provided by the equivalent voltage source, is the transient recovery voltage across the test circuit breaker, is the damping resistance, is the time delay capacitance.

[0019] Further, in step three, the power frequency recovery voltage peak value is taken as the voltage standard value, and the current corresponding to the power frequency recovery voltage peak value is taken as the current standard value, and the time domain is converted to the angle domain through dimensionless transformation, and the conversion relationship is:

[0020] ,

[0021] In the formula, represents angle, is the power frequency recovery voltage peak value, is the transient recovery voltage peak value, is the per unit value of the transient recovery voltage peak value, is the natural angular frequency of the undamped transient recovery voltage (TRV) in the test circuit; is the current corresponding to the power frequency recovery voltage peak value, which is the current standard value;

[0022] The dimensionless form of the equivalent model equation set is:

[0023] ,

[0024] And the following definitions are made:

[0025] ,

[0026] The dimensionless form of the equivalent model equation set is finally expressed as:

[0027] ,

[0028] In the formula, , The dimensionless parameters of capacitance and resistance are represented by and respectively, where represents the time delay capacitance unit value, represents the series damping unit value; , , and , , and corresponding unit values.

[0029] Further, in step four, according to the dimensionless form of the equivalent model equation set obtained in step three, the following differential equation is obtained:

[0030] ,

[0031] The initial conditions are:

[0032] ,

[0033] where, is the first-order derivative of with respect to , is the second-order derivative of with respect to .

[0034] Further, in step five: the dimensionless parameters of capacitance and resistance determined in step three are used as decision variables of the improved particle swarm optimization algorithm, i.e. the two dimensions of the particles; the initial parameters of the improved particle swarm optimization algorithm are set, including the number of particles, the number of iterations, the adaptive inertia weight, the adaptive learning factor, the error compensation mechanism, the particle initial position and the particle initial speed are randomized, and the upper and lower bounds of the particle position and speed are set.

[0035] Further, the initial parameters of the improved particle swarm optimization algorithm are set, including:

[0036] The state-aware variable step mechanism is adopted, which dynamically adjusts the search strategy by real-time monitoring of the population state, specifically:

[0037] In each iteration, the population diversity index and convergence rate factor are calculated, and their expressions are as follows:

[0038] ,

[0039] ,

[0040] in, Indicators representing diversity Represents the convergence rate factor. This represents the current iteration number. The number of particles, and , For the first The position of each particle. The optimal position globally. This represents the globally optimal fitness value for the current iteration step. This is the globally optimal fitness value from the previous iteration. The maximum value of the particle position. This represents the minimum position of the particle.

[0041] Adaptive inertia weights are obtained based on diversity indicators and convergence rate factors. The expression:

[0042] ,

[0043] In the formula, Based on linearly decreasing components, Indicates the maximum number of iterations. This represents the maximum value of the inertia weight. This represents the minimum value of the inertia weight. and For adjustment coefficients; when and When the population is determined to be trapped in a local optimum, the population size is increased. To enhance global exploration capabilities; when When it is determined to be in the rapid convergence period, reduce... To strengthen local development;

[0044] A multi-source error adaptive compensation mechanism is adopted to construct a personalized error compensation vector for each particle. The specific expression is:

[0045] ,

[0046] In the formula, This is a recent search trend compensation term, reflecting changes in the particle's own search direction. Indicates the first The individual optimal position of each particle in the current iteration step. Indicates the first The individual optimal position of each particle in the previous iteration step For the first The fitness value of each particle in the current iteration step. For the first The fitness value of each particle in the previous iteration step; This is a relative performance compensation term, based on the difference between the optimal performance of the particle and the population. As a historical memory compensation item, it preserves historical search experience, among which The attenuation coefficient is... For the first The error compensation vector for each particle in the previous iteration step; To compensate for the weighting, it is dynamically adjusted according to the search process, increasing during the search phase. and Increase during the development phase ;

[0047] An adaptive learning factor is adopted, including a cognitive learning factor and a social learning factor, which are dynamically adjusted according to the iteration progress and the population state. The expressions are as follows:

[0048] ,

[0049] ,

[0050] in, Represents cognitive learning factors, Represents social learning factors. , These are the initial and final values ​​of the cognitive learning factor, respectively. , These are the initial and final values ​​of the social learning factor, respectively. , To adjust the coefficients, the algorithm should focus more on individual cognition in the early stages and more on social learning in the later stages, while also fine-tuning the learning strategy based on population diversity.

[0051] Based on the aforementioned parameter settings, the particle velocity update formula is as follows:

[0052] ,

[0053] in, , for Random numbers within a range To compensate for the gain coefficient, For the first The velocity of each particle in the current iteration step For the first The velocity of each particle in the next iteration step;

[0054] The particle position update formula is:

[0055] ,

[0056] in For the first The position of each particle in the next iteration step.

[0057] Furthermore, step six, determining the specific steps for the improved particle swarm optimization algorithm's value function, includes:

[0058] Step 6.1 Solve the differential equation obtained in Step 4 using the ODE solver in MATLAB to obtain the per-unit voltage and angle across the circuit breaker. The waveform and the vertical and horizontal axis data;

[0059] Step 6.2 Find the maximum value in the y-axis data. ;

[0060] Step 6.3 Divide the ordinate of each data point by its abscissa to find the maximum value, and use the maximum value as the slope. It passes through the origin and has a slope of The straight line with a vertical coordinate value of The x-coordinate corresponding to time ,Right now: ; Calculate all data points before the data point corresponding to the maximum value of the y-coordinate up to this line. The distance is expressed as: Find the data point corresponding to the maximum distance value, and the data point that passes through this data point with a slope of [value missing]. The intersection of the straight line and the x-axis is denoted as . ,Right now: ,Will Divide by get , and These are the parameters that represent the waveform of the differential equation;

[0061] Step 6.4 Convert the dimensional standard values ​​for direct testing of high-voltage AC circuit breakers in GB / T 1984-2014 into dimensionless standard values, i.e., amplitude coefficients. and The result obtained in step 6.3 and The sum of the absolute values ​​of the differences between the individual values ​​is used as the fundamental value function for improving the particle swarm optimization algorithm, expressed as:

[0062] , a basic value function, and is a weight coefficient;

[0063] Step 6.5, in order to meet the standard, a hierarchical adaptive penalty mechanism is added in the value function, the hard constraint in the standard is converted into a hierarchical adaptive penalty with a soft constraint, and the hard constraint penalty term is:

[0064]

[0065] wherein, represents the hard constraint penalty term, is a parameter feasibility indication function, which takes a large value when the parameter exceeds the physically achievable range, and otherwise takes 0; , is a hard constraint term weight coefficient;

[0066] the soft constraint penalty term is:

[0067]

[0068] wherein, represents the soft constraint penalty term, is a parameter optimization function, which encourages the parameter configuration to be within the engineering preferred range, , is a soft constraint term weight coefficient; and is dynamically adjusted according to the constraint satisfaction, when significantly increases and reduces , and the default weight configuration is restored when the hard constraint is satisfied;

[0069] The final value function is represented as:

[0070] .

[0071] Further, an improved particle swarm optimization algorithm is run, and a dimensionless parameter combination corresponding to the minimum value function is obtained through iterative optimization, and actual inductance, capacitance and resistance values of the test circuit are obtained through formula conversion; Simulink simulation is performed based on the obtained inductance, capacitance and resistance values, and a corresponding transient recovery voltage (TRV) waveform is obtained, a peak value, a time to peak value and a time delay of the corresponding TRV waveform are extracted, and the effectiveness of the method is verified by comparing with the standard value.

[0072] The application also protects a computer device comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the parameter self-tuning method of the high-voltage alternating current circuit breaker short-circuit breaking test circuit when executing the computer program.​​

[0073] Compared with the prior art, the application has the beneficial effects that:

[0074] 1) High efficiency: the improved particle swarm optimization algorithm of the application has faster convergence speed and shorter running time, and can give results within a few seconds in multiple runs, which can greatly improve work efficiency and perform more circuit breaker test tasks in the same time compared with the experience debugging method and the standard particle swarm optimization algorithm;

[0075] 2) High precision: through the hierarchical adaptive penalty mechanism and the error compensation term, the algorithm can more accurately meet the engineering standard requirements, avoiding the problem of repeated debugging in traditional methods;

[0076] 3) Strong adaptability: this method can adapt to different test conditions, and only needs to change the standard parameter value to output parameters meeting the requirements in T100, T60, T30, T10 and other short-circuit test modes;

[0077] 4) Good robustness: the improved particle swarm optimization algorithm can dynamically adjust the inertia weight by real-time sensing of the convergence state and diversity of the population, balance the global exploration and local development capabilities, and has better stability and robustness;

[0078] 5) Expandability: this method combines differential equations and improved particle swarm optimization algorithm, and can be used not only in two-parameter test occasions, but also in four-parameter occasions such as synthetic test, only the expression of the differential equation and the value function of IPSO need to be changed. BRIEF DESCRIPTION OF DRAWINGS

[0079] Figure 1 is a schematic diagram of three-phase direct test circuit, wherein AC represents a three-phase alternating current source, MB represents a protective circuit breaker, represents a reactor, and T represents a three-phase winding transformer, , , composes a TRV frequency modulation device, and TB represents a test circuit breaker;

[0080] Figure 2 is a single-phase equivalent circuit model of the first open phase derived from Figure 1 , wherein represents an equivalent voltage source, represents an equivalent inductance, and is the transient recovery voltage across the test circuit breaker;

[0081] Figure 3 is a dimensionless parameter form of the single-phase equivalent circuit model of Figure 2 , wherein represents an equivalent voltage source, representative series damping unit, representative time delay capacitance unit;

[0082] Figure 4 Flow chart of working principle of improved particle swarm optimization algorithm provided by the present application;

[0083] Figure 5 Flow chart of generating waveform parameters of differential equation by the present application;

[0084] Figure 6 Schematic diagram of parameters to be extracted in TRV waveform specified by the standard;

[0085] Figure 7 (a) is the TRV waveform envelope parameter, and (b) is the dimensionless TRV waveform envelope parameter;

[0086] Figure 8 Simulink simulation waveforms of test circuit parameters obtained by the present application based on improved particle swarm optimization algorithm, under the rated voltage of 24kV and the rated short-circuit breaking current of 63kA, and under test modes T100, T60, T30 and T10, wherein (a) is the simulation waveform of T100, (b) is the simulation waveform of T60, (c) is the simulation waveform of T30, and (d) is the simulation waveform of T10;

[0087] Figure 9 Simulink simulation waveforms of test circuit parameters obtained by the present application based on improved particle swarm optimization algorithm, under the rated voltage of 40.5kV and the rated short-circuit breaking current of 63kA, and under test modes T100, T60, T30 and T10, wherein (a) is the simulation waveform of T100, (b) is the simulation waveform of T60, (c) is the simulation waveform of T30, and (d) is the simulation waveform of T10;

[0088] Figure 10 Simulink simulation waveforms of test circuit parameters obtained by the present application based on improved particle swarm optimization algorithm, under the rated voltage of 72.5kV and the rated short-circuit breaking current of 63kA, and under test modes T100, T60, T30 and T10, wherein (a) is the simulation waveform of T100, (b) is the simulation waveform of T60, (c) is the simulation waveform of T30, and (d) is the simulation waveform of T10;

[0089] Figure 11 Simulation model of the present application in Simulink. DETAILED DESCRIPTION

[0090] In order to make the technical solutions of the present application clearer, the technical solutions of the present application are described in further detail below in combination with the drawings and specific embodiments.

[0091] This invention provides a parameter self-tuning method for a short-circuit breaking test circuit of a high-voltage AC circuit breaker, used for parameter self-tuning of a three-phase direct short-circuit breaking test circuit of a high-voltage AC circuit breaker, such as... Figure 1 As shown, each phase of the three-phase direct test circuit includes a three-phase AC voltage source AC, a protective circuit breaker MB, a reactor L, a three-phase transformer T, a TRV frequency modulation device, and the circuit breaker under test TB. The TRV frequency modulation device includes a time delay branch and a frequency modulation branch connected in parallel. The time delay branch is a time delay capacitor. It is a separate component, with the FM branch consisting of FM capacitors. and damping resistor The series connection and parameter self-tuning method includes the following steps:

[0092] Step 1, such as Figure 2 As shown, based on symmetry, the three-phase direct test circuit is equivalent to a single-phase equivalent test circuit, with the phase containing the circuit breaker under test that is first broken being taken as the equivalent single phase. Simultaneously, the three-phase transformer T and reactor L in the three-phase direct test circuit are equivalent to inductors. The voltage source adopts an equivalent voltage source. Establish an equivalent model;

[0093] Step 2: Based on Kirchhoff's Current Law (KCL), Kirchhoff's Voltage Law (KVL), and the characteristics of passive components, write the time-domain equations and initial conditions of the equivalent model, specifically as follows:

[0094] ,

[0095] In the formula, Indicates time, and respectively flowing through the equivalent inductance and frequency modulation capacitor The current, The voltage provided by the equivalent voltage source. The transient recovery voltage across the circuit breaker under test. For damping resistor, For time delay capacitor;

[0096] Step 3: Select standard values ​​for voltage and current, transform the time domain to the angle domain, rewrite the equations and initial conditions, and obtain the equivalent model under dimensionless parameters (e.g., Figure 3 As shown), specifically:

[0097] Using the peak value of the power frequency recovery voltage as the voltage standard value and the current corresponding to the peak value of the power frequency recovery voltage as the current standard value, a per-unit transformation is performed, and the time domain is converted to the angle domain, such as... Figure 6 , Figure 7The conversion relationship is shown in (a) and (b):

[0098] ,

[0099] In the formula, represents an angle, is the peak value of the power frequency recovery voltage, is the peak value of the transient recovery voltage, is the peak value of the transient recovery voltage, is the natural angular frequency of the undamped transient recovery voltage (TRV) in the test circuit; is the peak value of the power frequency recovery voltage corresponding to the current, and is the current standard value;

[0100] The dimensionless form of the equivalent model equation set is:

[0101] ,

[0102] And the following definitions are made:

[0103] ,

[0104] The dimensionless form of the equivalent model equation set is finally expressed as:

[0105] ,

[0106] In the formula, , represent the dimensionless parameters of the capacitance and resistance, respectively, wherein represents the time delay capacitance unit value, represents the series damping unit value; , , are the unit values corresponding to , and respectively;

[0107] Step four: According to the equation set obtained in step three, the differential equation relationship between the unit value of the voltage across the tested circuit breaker and other dimensionless parameters is arranged, and the final result obtained is:

[0108] ,

[0109] The initial conditions are:

[0110] ,

[0111] Among them, is the first-order derivative of with respect to , and is with respect to second derivative of

[0112] Step five, set the initial parameters of IPSO, including the number of particles, the number of iterations, the adaptive inertia weight, the adaptive learning factor, the error compensation mechanism, randomize the initial position and initial velocity of particles, and set the upper and lower bounds of particle position and velocity, as shown in Figure 4 , specifically:

[0113] The number of particles is set to 50, and the dimension of each particle is 2, which is , Since the decision variable is dimensionless, the range of is (0, 1), the range of is (0, 0.5); use the rand function to randomly allocate and of each particle, and also randomly set the velocity; the maximum number of iterations is set to 200, indicates the current number of iterations.

[0114] Adopt state-aware variable step mechanism, dynamically adjust the search strategy by real-time monitoring of population state, calculate population diversity index and convergence speed factor in each iteration, their expressions are respectively:

[0115] ,

[0116] ,

[0117] Where, represents the diversity index, represents the convergence speed factor, is the current iteration number, is the number of particles, and , is the position of the th particle, is the global optimal position, is the global optimal fitness value of the current iteration step, is the global optimal fitness value of the last iteration step, is the maximum value of particle position, is the minimum value of particle position.

[0118] Based on the diversity index and the convergence speed factor, the expression of the adaptive inertia weight is obtained:

[0119] ,

[0120] wherein, is the base linear decreasing component, denotes the maximum number of iterations, is the inertia weight maximum value, is the inertia weight minimum value; and is the adjustment coefficient; when and , it is determined that the population falls into local optimum, at this time, the is increased to enhance the global exploration ability; when , it is determined that it is in the rapid convergence period, the is reduced to strengthen the local development;

[0121] A multi-source error adaptive compensation mechanism is adopted to construct a personalized error compensation vector for each particle, and the specific expression is:

[0122] ,

[0123] wherein, is a recent search trend compensation term, reflecting the change of the search direction of the particle itself, wherein denotes the individual best position of the th particle in the current iteration step, denotes the individual best position of the th particle in the last iteration step, is the fitness value of the th particle in the current iteration step, is the fitness value of the th particle in the last iteration step; is a relative performance compensation term, based on the gap between the particle and the group optimum; is a historical memory compensation term, which retains historical search experience, wherein is a decay coefficient, is the error compensation vector of the th particle in the last iteration step; is a compensation weight, which is dynamically adjusted according to the search process, and is increased in the search stage and , and is increased in the development stage ;

[0124] An adaptive learning factor is adopted, including a cognitive learning factor and a social learning factor, which is dynamically adjusted according to the iteration progress and the population state, and the expressions are respectively:

[0125] ,

[0126] ,

[0127] in, Represents cognitive learning factors. Represents social learning factors. , These are the initial and final values ​​of the cognitive learning factor, respectively. , These are the initial and final values ​​of the social learning factor, respectively. , To adjust the coefficients, the algorithm should focus more on individual cognition in the early stages and more on social learning in the later stages, while also fine-tuning the learning strategy based on population diversity.

[0128] Based on the aforementioned parameter settings, the particle velocity update formula is as follows:

[0129] ,

[0130] in, , for Random numbers within a range To compensate for the gain coefficient, For the first The velocity of each particle in the current iteration step For the first The velocity of each particle in the next iteration step;

[0131] The particle position update formula is:

[0132] ,

[0133] in For the first The position of each particle in the next iteration step.

[0134] Step Six: Based on the differential equation relationship obtained in Step Four, determine the value function of the improved particle swarm optimization algorithm, such as... Figure 5 As shown, it includes the following steps:

[0135] Step 6.1 Solve the differential equation obtained in Step 4 using the ODE solver in MATLAB to obtain the per-unit voltage and angle across the circuit breaker. The waveform and its vertical and horizontal coordinate data;

[0136] Step 6.2 Find the maximum value in the y-axis data. ;

[0137] Step 6.3 Divide the ordinate of each data point by its abscissa to find the maximum value, and use the maximum value as the slope. It passes through the origin and has a slope of the straight line with slope and passing through the data point with the maximum value of the ordinate , i.e. , and the distance between all data points before the data point with the maximum value of the ordinate and the straight line is calculated, denoted as , the data point corresponding to the maximum distance value is found, and the intersection point of the straight line with slope passing through the data point and the abscissa is denoted as , i.e. , and is divided by to obtain , and are parameters representing the waveform of the differential equation;

[0138] Step 6.4. Convert the dimension standard values related to the direct test of high-voltage alternating current circuit breakers in GB_T 1984-2014 into dimensionless standard values, i.e. amplitude coefficients and , and take the absolute value of the difference between and obtained in step 6.3, as the basic value function of the improved particle swarm optimization algorithm, denoted as

[0139] ,

[0140] In the formula, denotes the basic value function, and are weight coefficients;

[0141] Step 6.5. To meet the requirements of the standard, add a hierarchical adaptive penalty mechanism to the value function, convert the hard constraints and soft constraints in the standard into hierarchical adaptive penalties, and the hard constraint penalty term is

[0142] ,

[0143] wherein, denotes the hard constraint penalty term, is a parameter feasibility indicator function that takes a large value when the parameter exceeds the physically achievable range, and otherwise takes 0; , is the hard constraint term weight coefficient;

[0144] The soft constraint penalty term is

[0145] ,

[0146] wherein, denotes the soft constraint penalty term, For parameter optimization function, encourage parameter configuration in engineering optimization range, 、 Soft constraint term weight coefficient; weight coefficient and According to the constraint satisfaction, dynamic adjustment, when Significant increase And reduce , when the hard constraint is satisfied, the default weight configuration is restored;

[0147] The final value function Indicated as:

[0148] .

[0149] Step seven, run the improved particle swarm optimization algorithm, iterative optimization to get the dimensionless parameter combination corresponding to the minimum value function, through the formula conversion to get the actual inductance, capacitance and resistance value of the test circuit; Based on the obtained to inductance, capacitance and resistance value, Simulink simulation is carried out, and the corresponding transient recovery voltage (TRV) waveform is obtained. Extract the peak value, peak value time and time delay of the corresponding TRV waveform, and compare them with the standard value to verify the effectiveness of the method:

[0150] The conversion relationship between dimensionless parameter and dimensional parameter is:

[0151] ,

[0152] In the formula, The first opening coefficient, The rated voltage of the test circuit breaker, The breaking current of the test circuit breaker, , , The angle corresponding to the vertical coordinate 1 in the dimensionless TRV envelope line;

[0153] After conversion, the inductance, capacitance and resistance values are input into the simulation model shown in Figure 11 Run the simulation to get the TRV waveform curve and Compare them with the standard value (as shown in Table 3) to verify the effectiveness of the method:

[0154] Under the conditions of rated short-circuit breaking current 63kA and rated voltage 24kV, 40.5kV and 72.5kV, T100, T60, T30 and T10 four kinds of short-circuit test methods were tested. Table 1 shows the optimal dimensionless parameters obtained by iterative optimization of improved particle swarm optimization algorithm The relevant test circuit parameters obtained by conversion 、 , and ; Figure 8 to Figure 10 The corresponding TRV curves obtained from simulations based on algorithm-optimized parameters are presented: Figure 8 , Figure 9 , Figure 10 The simulated TRV waveforms for four short-circuit test modes are presented under the conditions of a rated short-circuit breaking current of 63kA and rated voltages of 24kV, 40.5kV, and 72.5kV, respectively. In each figure, (a), (b), (c), and (d) correspond to the simulated waveforms of short-circuit test modes T100, T60, T30, and T10, respectively. Table 2 shows the peak values ​​of the corresponding TRV curves. Delay Time to reach peak The value, compared with the standard values ​​shown in Table 3, shows that the delay... Relatively small, peak value Time to reach peak The results closely match the standard values, demonstrating the effectiveness of the method of this invention.

[0155] Table 1

[0156]

[0157] Table 2

[0158]

[0159] Table 3

[0160]

[0161] In one specific embodiment, the present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the parameter self-tuning method for the short-circuit breaking test circuit of the high-voltage AC circuit breaker.

[0162] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Any modifications, equivalent substitutions, or improvements made by those skilled in the art within the scope of the technology disclosed in this invention, based on the technical solution and concept of the present invention, should be included within the protection scope of this invention. Therefore, the protection scope of this invention should be determined by the scope of the claims.

Claims

1. A parameter self-tuning method for short-circuit breaking test circuit of high-voltage AC circuit breaker, which is used for parameter self-tuning of three-phase direct test circuit of short-circuit breaking of high-voltage AC circuit breaker; each phase of the three-phase direct test circuit comprises a voltage source, a protection circuit breaker, a reactor, a transformer, a TRV frequency modulation device and a circuit breaker under test, the TRV frequency modulation device comprises a time delay branch and a frequency modulation branch connected in parallel, the time delay branch is composed of a time delay capacitor alone, and the frequency modulation branch is composed of a frequency modulation capacitor and a damping resistor connected in series, characterized in that, The parameter self-tuning method comprises the following steps: Step one, according to symmetry, the three-phase direct test circuit is equivalent to a single-phase equivalent test circuit, taking the phase where the first tripped circuit breaker is located as the equivalent single phase, and the transformer and the reactor in the three-phase direct test circuit are equivalent to an inductor, and an equivalent voltage source is used to establish an equivalent model; Step two, according to Kirchhoff's current law, Kirchhoff's voltage law and the characteristics of passive elements, time-domain equations and initial conditions of the equivalent model are written; Step three, standard values of voltage and current are selected for dimensionless transformation, and the time domain is converted to the angle domain, and the equations and initial conditions are written again to obtain the equivalent model under dimensionless parameters; Step four, according to the equations obtained in step three, a differential equation relationship between the dimensionless voltage across the circuit breaker under test and other dimensionless parameters is obtained; Step five, the dimensionless parameters obtained in step three are used as decision variables of the improved particle swarm optimization algorithm, and initial parameters of the improved particle swarm optimization algorithm are set; Step six, according to the differential equation relationship obtained in step four, a value function of the improved particle swarm optimization algorithm is determined, and the value function adopts a hierarchical adaptive penalty mechanism; Step seven, the improved particle swarm optimization algorithm is iteratively run to find a dimensionless parameter combination that minimizes the value function, and actual inductance, capacitance and resistance values of the test circuit are obtained through formula conversion.

2. The parameter self-tuning method for a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 1, characterized in that, The time-domain equations and initial conditions of the equivalent model written in step two are: ， wherein denotes time, and are the currents flowing through the equivalent inductances and the frequency-modulated capacitance , is the voltage provided by the equivalent voltage source, is the transient recovery voltage across the tested circuit breaker, is the damping resistance, is the time-delay capacitance.

3. The parameter self-tuning method for a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 2, characterized in that, In step three, the peak value of the power frequency recovery voltage is taken as the voltage standard value, and the current corresponding to the peak value of the power frequency recovery voltage is taken as the current standard value for dimensionless transformation, and the time domain is converted to the angle domain, and the conversion relationship is: ， wherein denotes the angle, is the peak value of the power frequency recovery voltage, is the peak value of the transient recovery voltage, is the peak value of the transient recovery voltage, is the natural angular frequency of the undamped transient recovery voltage in the test circuit; is the peak value of the power frequency recovery voltage, and is the current standard value. And the following definitions are made: ， Then the dimensionless form of the equivalent model equation set is finally represented as: , wherein , respectively represent the dimensionless parameters of capacitance and resistance, wherein represents the time delay capacitance dimensionless value, represents the series damping dimensionless value; , , , are respectively , , and the corresponding dimensionless values.

4. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 3, characterized in that, In step four, according to the dimensionless form of the equivalent model equation set obtained in step three, the following differential equation is obtained: ， The initial conditions are: ， wherein is relative to the first derivative of is relative to the second derivative of 5. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 4, characterized in that, The dimensionless parameters of the capacitance and resistance determined in step three are used as decision variables of the improved particle swarm optimization algorithm, i.e. two dimensions of particles; initial parameters of the improved particle swarm optimization algorithm are set, including the number of particles, the number of iterations, the adaptive inertia weight, the adaptive learning factor, the error compensation mechanism, the initial position and the initial speed of the particles are randomized, and the upper and lower bounds of the particle position and speed are set.

6. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 5, characterized in that, The improved particle swarm algorithm adopts a state-aware variable step mechanism, which dynamically adjusts the search strategy by monitoring the population state in real time, specifically: In each iteration, the population diversity index and the convergence speed factor are calculated, and their expressions are respectively: ， ， in, Indicators representing diversity Represents the convergence rate factor. This represents the current iteration number. The number of particles, and , For the first The position of each particle. The optimal position globally. This represents the globally optimal fitness value for the current iteration step. This is the globally optimal fitness value from the previous iteration. The maximum value of the particle position. This represents the minimum position of the particle. An expression for the adaptive inertia weight based on diversity index and convergence speed factor is obtained. ， wherein, is the base linear decreasing component, represents the maximum iteration number, is the inertia weight maximum value, is the inertia weight minimum value; and is the adjustment coefficient; when and the population falls into local optimum, at this time, increase to enhance the global exploration ability; when is in the fast convergence period, reduce to strengthen the local development.

7. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 6, characterized in that, A multi-source error adaptive compensation mechanism is adopted to construct a personalized error compensation vector for each particle The specific expression is: ， In the formula, This is a recent search trend compensation term, reflecting changes in the particle's own search direction. Indicates the first The individual optimal position of each particle in the current iteration step. Indicates the first The individual optimal position of each particle in the previous iteration step For the first The fitness value of each particle in the current iteration step. For the first The fitness value of each particle in the previous iteration step; This is a relative performance compensation term, based on the difference between the optimal performance of the particle and the population. As a historical memory compensation item, it preserves historical search experience, among which The attenuation coefficient is... For the first The error compensation vector for each particle in the previous iteration step; To compensate for the weighting, it is dynamically adjusted according to the search process, increasing during the search phase. and Increase during the development phase ; An adaptive learning factor is adopted, including a cognitive learning factor and a social learning factor, which are dynamically adjusted according to the iteration progress and the population state, and their expressions are respectively: , , wherein, represents a cognitive learning factor, represents a social learning factor, , are respectively an initial value and a final value of the cognitive learning factor, , are respectively an initial value and a final value of the social learning factor, , is an adjustment coefficient; Based on the parameter settings, the particle speed update formula is: , in, , for Random numbers within a range To compensate for the gain coefficient, For the first The velocity of each particle in the current iteration step For the first The velocity of each particle in the next iteration step; The particle position update formula is: ， wherein is the position of the th particle at the next iteration step.

8. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 7, characterized in that, The specific steps of step six for determining the value function of the improved particle swarm optimization algorithm include: Step 6.1 Solving the differential equation obtained in Step 4 by ODE solver in MATLAB to get the waveform of the voltage in per unit and angle at the terminals of the circuit breaker and the data of the longitudinal and transverse coordinates ; Step 6.2 Find the maximum value in the ordinate data ; Step 6.3 Divide the ordinate of each data point by the abscissa, find the maximum value, and take the maximum value as the slope The straight line passing through the origin and having a slope of corresponds to the abscissa when the ordinate value is , that is: ; the distance from all data points before the data point corresponding to the maximum value of the ordinate to the straight line is calculated and expressed as: The data point corresponding to the maximum distance value is found, and the intersection of the straight line passing through this data point and having a slope of with the abscissa is denoted as , that is: ; and is divided by to obtain , and are parameters representing the waveform of the differential equation. Step 6.4 Convert the dimensioned standard values of the direct tests of the high voltage AC circuit breaker into dimensionless standard values, i.e. amplitude factors and Step 6.3 and The sum of the absolute values of the differences between each of these is taken as the basis for the merit function of the improved particle swarm optimization algorithm, and is denoted as: , represents a base value function, and are weight coefficients; Step 6.5 To meet the standards, a hierarchical adaptive penalty mechanism is added to the value function to convert the hard constraints and soft constraints in the standards into hierarchical adaptive penalties, and the hard constraint penalty term is: ， wherein, represents a hard constraint penalty term, is a parameter feasibility indicator function that takes a large value when the parameter is outside the physically achievable range, and 0 otherwise; , is a hard constraint term weight coefficient; The soft constraint penalty term is: ， wherein, represents a soft constraint penalty term, is a parameter preference function, encouraging parameter configurations within an engineering preference range, , is a soft constraint term weight coefficient; and is dynamically adjusted according to constraint satisfaction, when significantly increases and decreases , and returns to default weight configuration when hard constraints are satisfied; Final value function is represented as: 。 9. The parameter self-adjusting method of a short-circuit breaking test circuit of a high-voltage AC circuit breaker according to claim 8, characterized in that, The improved particle swarm optimization algorithm is run, and a dimensionless parameter combination corresponding to a minimum value function is obtained through iterative optimization; actual inductance, capacitance and resistance values of the test circuit are obtained through formula conversion; Simulink simulation is carried out based on the obtained inductance, capacitance and resistance values, and corresponding transient recovery voltage waveforms are obtained; the peak value, time of reaching the peak value and time delay of the corresponding transient recovery voltage waveforms are extracted, and compared with standard values, so as to verify the effectiveness of the method.

10. A computer apparatus comprising a memory and a processor, characterised in that, The computer program is stored on the memory, and the processor executes the computer program to realize the parameter self-tuning method of the high-voltage alternating-current circuit breaker short-circuit breaking test circuit according to any one of claims 1 to 9.

Images