Method for detecting instantaneous voltage drop of series filter

Abstract

The invention discloses a series filter voltage instantaneous drop detection method. The method specifically comprises the following steps: (1) obtaining a distortion voltage value of a three-phase voltage based on a voltage compensation instruction of an instantaneous reactive power dq0 method; and (2) taking the group of distorted voltage values as input values to enter a PI controller, performing PI parameter optimization by the PI controller based on an improved immune genetic algorithm, and performing detection and voltage compensation by driving an inverter bridge after a PWM signal is generated. The method has the advantages that decoupling control is conducted on the cross coupling amount between the dq axis of the series filter, the control performance is improved, the improved immune genetic algorithm is adopted to optimize the controller parameters, the problems of blindness and randomness of searching of a traditional genetic algorithm are solved, meanwhile, the PI controller parameter optimization speed is high in response, robustness is high, and the method is suitable for large-scale popularization and application. The problem that controller parameters are difficult to adapt to changing environments is solved, and the voltage detection precision and compensation performance of the series filter are improved.

Description

Technical Field

[0001] The invention relates to the technical field related to power distribution networks, and in particular to a method for detecting instantaneous voltage drop of a series filter. Background Art

[0002] In recent years, renewable energy sources such as photovoltaics and wind power have developed rapidly. Due to the intermittent, complex, and instability characteristics of renewable energy power generation, its output power has large fluctuations. As the proportion of access to the power grid increases, its power quality problems become more and more prominent. Voltage drop has become one of the prominent power quality problems in distribution networks. Series filters can effectively compensate for voltage drop and other voltage quality problems. They are important management equipment for solving power quality problems in distribution networks. Therefore, their voltage drop detection and control methods have received widespread attention from researchers.

[0003] At present, the voltage drop detection methods are mainly divided into the following three categories: (1) αβ0 transformation method based on instantaneous reactive power theory. This method is only applicable to the case where the system voltage is three-phase sinusoidal symmetry, otherwise there will be a relatively large detection error. (2) dq0 transformation method. This method can be applied to any non-sinusoidal and asymmetric three-phase circuit. (3) Wavelet analysis method. The huge amount of computation of the wavelet transform method leads to poor real-time performance, which has become an important factor restricting the field application of wavelet transform. Comparing the above three main voltage drop detection methods, the dq0 transformation method is the most widely studied. In the dq0 transformation method, PI control can effectively detect and compensate the voltage, but if the control parameters are not set properly, it will directly affect the voltage drop detection effect of the series filter. The traditional PI control method mainly performs empirical trial and error, but it is easy to fall into local optimality, and when the fixed parameters are controlled by the external environment, its dynamic response performance and adaptability will be reduced, and it cannot meet the requirements of high-performance and high-precision voltage detection. However, microgrids are easily affected by the environment, which can easily cause output power fluctuations at the distribution network's connection points, resulting in voltage drops at the connection points. For traditional PI controllers, when the absolute value of the voltage error is large, the parameter control response characteristics deteriorate, making it difficult to apply to power quality control of distribution networks containing microgrids.

[0004] In order to effectively improve the performance of traditional PI controllers, there are many methods for optimizing PI controller parameters, including fuzzy control, neural network, genetic method, etc. Both fuzzy controllers and neural networks have been proven to be universal function approximators. Genetic algorithms belong to global optimization algorithms, which optimize the performance of control systems by obtaining a set of optimal control parameters. Due to the randomness of search in genetic algorithms, blind search is likely to occur, which slows down the convergence speed and deteriorates the convergence stability. Immune genetic algorithm is an intelligent optimization method that introduces biological immune mechanism based on traditional genetic algorithm. It can better solve the degradation phenomenon of genetic algorithm, but it also has the blindness and randomness of search. Summary of the invention

[0005] The present invention aims to overcome the above-mentioned deficiencies in the prior art and provides a method for detecting instantaneous voltage drop of a series filter, which can improve the detection accuracy and compensation performance of the series filter voltage.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for detecting instantaneous voltage drop of a series filter, specifically comprising:

[0008] (1) The distorted voltage value obtained by the voltage compensation instruction of the three-phase voltage based on the instantaneous reactive power dq0 method;

[0009] (2) The set of distorted voltage values ​​is used as input values ​​to enter the PI controller. The PI controller optimizes the PI parameters based on the improved immune genetic algorithm. After the PWM signal is generated, it drives the inverter bridge to detect and compensate the voltage.

[0010] In view of the above problems, the present invention proposes a series filter voltage drop detection algorithm based on an improved immune genetic algorithm. The core of the algorithm is to decouple the cross-coupling amount between the dq axes of the series filter to improve the control performance, and use the improved immune genetic algorithm to optimize the controller parameters. While solving the blindness and randomness problems of traditional genetic algorithm search, the PI controller parameter optimization speed is fast and robust, which solves the problem that the controller parameters are difficult to adapt to the changing environment, and improves the voltage detection accuracy and compensation performance of the series filter.

[0011] Preferably, in step (1), the operation of obtaining the distorted voltage value is as follows:

[0012] Assume that the three-phase voltage expression in the distribution network system is as shown in formula (1):

[0013]

[0014] In the formula, u a ,u b ,u c is the three-phase voltage of the distribution network, is the three-phase positive sequence component, as shown in formula (2):

[0015]

[0016] Where U + Refers to the maximum forward voltage, that is, the forward amplitude. Conversely, U - is the negative amplitude. After the introduction of the nth harmonic, it is expressed as the positive and negative amplitude of the nth harmonic; ω refers to the angular frequency; ψ refers to the initial phase angle, which is the same as U and can be positive or negative. After the introduction of the nth harmonic, it is expressed as the initial phase angle of the nth harmonic; t refers to the time variable, is the three-phase negative sequence component, as shown in formula (3):

[0017]

[0018] Where u0 is the zero-sequence component, as shown in equation (4):

[0019]

[0020] In the formula, is the positive sequence component of the nth harmonic, as shown in formula (5):

[0021]

[0022] In the formula, is the negative sequence component of the nth harmonic, as shown in equation (6):

[0023] Three-phase voltage u dq0 and u abc The transformation relationship between them is obtained by the following formula (7):

[0024]

[0025] In the formula

[0026]

[0027] In the formula, The vector representing the positive sequence component Park;

[0028] A vector representing the negative sequence component park;

[0029] u 00 A vector representing the zero-sequence component.

[0030] Filter out the high-order harmonics from equation (7) with a low-pass filter, and get the DC component expression (8):

[0031]

[0032] Performing dq0 inverse transformation on equation (8), we can obtain the fundamental voltage positive sequence component: As shown in formula (9):

[0033]

[0034] The instantaneous drop voltage detected by the series filter is shown in formula (10):

[0035]

[0036] Preferably, in step (2), the mathematical model of the PI controller is formula (11):

[0037]

[0038] Where K is the proportionality coefficient, K p is the voltage compensation parameter, is the integral time constant, e(t) is the deviation value of the PI controller input at time t; u(t) is the output value of the PI controller at time t; t is the time variable; in the actual digital control field, digital PI control is often used, using the summation form to replace the integral and the increment form to replace the differential. The approximate transformation is as follows:

[0039]

[0040] Where, T s is the sampling period, k is the sampling sequence, i is the variable parameter, and L represents the total number of sampling sequences. The discrete PI control expression obtained by simplifying e(kT) is as shown in formula (13):

[0041]

[0042] In the above formula, u(k) is the output value at the kth sampling time, e(k) is the deviation value input at the kth sampling time, K I is the integration coefficient, u0 is the original initial value for starting the PI controller.

[0043] Preferably, in step (2), the specific principle of the PI controller parameter optimization method based on the improved immune genetic algorithm is:

[0044] Will The parameter optimization sets an optimization objective function F, as shown in formula (14):

[0045]

[0046] Where T is the sampling time, T s is the number of sampling points, MinF represents the minimum value of the objective function F, e(t i ) is t i The deviation value input at the sampling time, i is the parameter variable, t i =iT S .

[0047] Preferably, the PI parameter optimization steps based on the improved immune genetic algorithm are as follows:

[0048] (21) Select voltage compensation parameter K P , K IAs the antigen Ag, using the forward constraint, N K I parameter variable, and K I (N+1) is set as antibody Ab;

[0049] (22) Selection, crossover, mutation, vaccination; Calculate the fitness according to the PI performance optimization evaluation index of formula (14), iterate using formula (13), if the iteration termination condition is met, the algorithm stops and outputs the optimal result, otherwise repeat this step;

[0050] (23) Generate the next generation of immune antibodies and determine whether the conditions are met; if so, stop and output the optimal result. In this way, the parameter K of the objective function takes the minimum value under the premise of satisfying the system stability. P (N+1),K I (N+1) is the global optimal solution K of the final PI parameter P , K I , otherwise return to step (2) and restart the fitness calculation.

[0051] Preferably, in step (22) and step (23), for each antigen Ag, the operation is as follows:

[0052] (a) Option: Calculate each antibody Ab separately i and the same antigen Ag i The affinity coefficient β ij , directly call the shortest Euclidean distance formula to calculate:

[0053]

[0054] Where i represents the i-th generation, j represents the j-th, n represents the total number of antigens in the sample, and k = 1, 2, 3, ... n;

[0055] (b) Crossover: According to β ij Screen and sort the antibodies, select the m antibodies with the highest affinity as network cells, and clone the m network cells;

[0056] (c) Mutation: The cloned cells are subjected to mutation operation using the cell mutation formula C=C-α(CX), where C represents cloned antibody cells, X represents cloned antigen cells, and α represents the mutation rate;

[0057] (d) Immunization: recalculate the affinity of C after the mutation operation, and select the one with the highest affinity C=60 as the partial memory cell dataset Mp;

[0058] (e) Generate the next generation of new immune antibodies: Calculate the similarity S between each antibody ij , eliminate the similarity S in Mp ijGreater than the threshold σ s Individuals:

[0059]

[0060] (f) Merge Mp into the memory dataset M and update the dataset M.

[0061] Preferably, in step (23), specifically: select from the N individuals (0, 1) generated in step (21), replace the individuals with poor affinity, and use the memory data set M obtained from the previous immune calculation as antibodies for the next generation of immune calculation, and return to step (22), calculate the fitness value of the population, and observe whether the termination condition is met.

[0062] The beneficial effects of the present invention are: decoupling control is performed on the cross-coupling amount between the dq axes of the series filter to improve the control performance, and the controller parameters are optimized using an improved immune genetic algorithm. While solving the blindness and randomness problems of the traditional genetic algorithm search, the PI controller parameter optimization speed has a fast response and strong robustness, and the problem that the controller parameters are difficult to adapt to the changing environment is solved, and the voltage detection accuracy and compensation performance of the series filter are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 It is a block diagram of PI controller parameter optimization in the present invention;

[0064] Figure 2 This is a diagram of a method for obtaining voltage compensation instructions based on the instantaneous reactive power dq0 method in the present invention. DETAILED DESCRIPTION

[0065] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments.

[0066] like Figure 1 In the embodiment described, a method for detecting instantaneous voltage drop of a series filter is specifically as follows:

[0067] (1) The three-phase voltage is obtained by the voltage compensation instruction based on the instantaneous reactive power dq0 method to obtain the distorted voltage value; the operation of obtaining the distorted voltage value is as follows:

[0068] Figure 2 In the formula (1), LPF (Low pass filter) is a low pass filter. Assume that the three-phase voltage expression in the distribution network system is as shown in formula (1):

[0069]

[0070] In the formula, u a ,u b ,u c is the three-phase voltage of the distribution network, is the three-phase positive sequence component, as shown in formula (2):

[0071]

[0072] Where U + Refers to the maximum forward voltage, that is, the forward amplitude. Conversely, U - is the negative amplitude. After the introduction of the nth harmonic, it is expressed as the positive and negative amplitude of the nth harmonic; ω refers to the angular frequency; ψ refers to the initial phase angle, which is the same as U and can be positive or negative. After the introduction of the nth harmonic, it is expressed as the initial phase angle of the nth harmonic; t refers to the time variable, is the three-phase negative sequence component, as shown in formula (3):

[0073]

[0074] Where u0 is the zero-sequence component, as shown in equation (4):

[0075]

[0076] In the formula, is the positive sequence component of the nth harmonic, as shown in formula (5):

[0077]

[0078] In the formula, is the negative sequence component of the nth harmonic, as shown in equation (6):

[0079]

[0080] Three-phase voltage u dq0 and u abc The transformation relationship between them is obtained by the following formula (7):

[0081]

[0082] In the formula

[0083]

[0084] In the formula, The vector representing the positive sequence component Park;

[0085] A vector representing the negative sequence component park; |

[0086] u 00 A vector representing the zero-sequence component.

[0087] Filter out the high-order harmonics from equation (7) with a low-pass filter, and get the DC component expression (8):

[0088]

[0089] Performing dq0 inverse transformation on equation (8), we can obtain the fundamental voltage positive sequence component: As shown in formula (9):

[0090]

[0091] The instantaneous drop voltage detected by the series filter is shown in formula (10):

[0092]

[0093] (2) The set of distorted voltage values ​​is used as input values ​​to enter the PI controller, and the PI controller optimizes the PI parameters based on the improved immune genetic algorithm. After the PWM signal is generated, it drives the inverter bridge to detect and compensate the voltage;

[0094] The mathematical model of the PI controller is as follows:

[0095]

[0096] Where K is the proportionality coefficient, K p is the voltage compensation parameter, is the integral time constant, e(t) is the deviation value of the PI controller input at time t; u(t) is the output value of the PI controller at time t; t is the time variable; in the actual digital control field, digital PI control is often used, using the summation form to replace the integral and the increment form to replace the differential. The approximate transformation is as follows:

[0097]

[0098] Where, T s is the sampling period, k is the sampling sequence, i is the variable parameter, and L represents the total number of sampling sequences. The discrete PI control expression obtained by simplifying e(kT) is as shown in formula (13):

[0099]

[0100] In the above formula, u(k) is the output value at the kth sampling time, e(k) is the deviation value input at the kth sampling time, and K I is the integration coefficient, u0 is the original initial value for starting the PI controller.

[0101] From the above analysis, it can be seen that the use of PI control can effectively detect and compensate voltage, but if the control parameter K of the PI controller is improperly set, it will directly affect the voltage drop detection effect of the series filter. At present, the optimization of the control parameters of the PI controller is mainly based on empirical trial and error, but it is easy to fall into local optimality, and when the fixed setting parameters are affected by changes in the controlled external environment, its dynamic response performance and adaptability will be reduced, and it cannot meet the requirements of high-performance and high-precision voltage detection. If the traditional immune genetic algorithm is used, although it can solve the problem that fixed parameters cannot adapt to environmental changes, it is prone to blindness and randomness. Therefore, the present invention proposes a PI controller parameter optimization method based on an improved immune genetic algorithm. The specific principle is,

[0102] Will The parameter optimization sets an optimization objective function F, as shown in formula (14):

[0103]

[0104] Where T is the sampling time, T s is the number of sampling points, MinF represents the minimum value of the objective function F, e(t i ) is t i The deviation value input at the sampling time, i is the parameter variable, t i =iT S .

[0105] according to Figure 1 As known from formula (10), the distorted voltage value u obtained by the dq0 method is ac ,u bc ,u cc , this group of voltage values ​​enters the PI controller as input values, and the PI parameter optimization steps based on the improved immune genetic algorithm are as follows:

[0106] (21) Select voltage compensation parameter K P , K I As the antigen Ag, using the forward constraint, N K I parameter variable, and K I (N+1) is set as antibody Ab; here we assume that K P The value range of K is [0, 50]. I The value range of is [0, 1]. In order to reduce the search amount, the number of iterations is set to be less than 100, that is, the maximum value of N is 100, and the control coefficient μ=2.

[0107] (22) Selection, crossover, mutation, and immunization; Calculate the fitness according to the PI performance optimization evaluation index of formula (14), and iterate using formula (13). If the iteration termination condition is met, the algorithm stops and outputs the optimal result, otherwise repeat this step. For each antigen Ag, the operation is as follows:

[0108] (a) Option: Calculate each antibody Ab separately i and the same antigen Ag i Affinity coefficient (coupling coefficient) β ij , directly call the shortest Euclidean distance formula to calculate:

[0109]

[0110] Where i represents the i-th generation, j represents the j-th, n represents the total number of antigens in the sample, k = 1, 2, 3, ... n;

[0111] (b) Crossover: According to β ij Screen and sort the antibodies, select the m antibodies with the highest affinity as network cells, and clone the m network cells;

[0112] (c) Mutation: The cloned cells are subjected to mutation operation using the cell mutation formula C = C-α(CX), where C represents cloned antibody cells, X represents cloned antigen cells, and a represents the mutation rate, which is generally 0.1.

[0113] (d) Immunization: recalculate the affinity of C after the mutation operation, and select the one with the highest affinity C=60 as the partial memory cell dataset Mp;

[0114] (e) Generate the next generation of new immune antibodies: Calculate the similarity S between each antibody ij , eliminate the similarity S of σ in Mp ij Greater than the threshold σ s Individuals:

[0115]

[0116] (f) Merge Mp into the memory dataset M and update the dataset M.

[0117] (23) After steps (e) and (f), the next generation of immune antibodies is generated, and it is determined whether the conditions are met. Select from the N individuals (0, 1) generated in step (21), replace the individuals with poor affinity, and use them together with the memory data set M obtained from the previous immune calculation as the antibodies for the next generation of immune calculations, and return to step (22) to calculate the fitness value of the population to observe whether the end condition is met. If it is met, stop and output the optimal result. In this way, the parameter K of the objective function takes the minimum value under the premise of satisfying the stability of the system.P (N+1), K I (N+1) is the global optimal solution K of the final PI parameter P , K I , otherwise return to step (2) and restart the fitness calculation.

[0118] In summary, the present invention proposes a series filter voltage drop detection algorithm based on an improved immune genetic algorithm. The core of the algorithm is to decouple the cross-coupling amount between the dq axes of the series filter to improve the control performance, and use the improved immune genetic algorithm to optimize the controller parameters. While solving the blindness and randomness problems of traditional genetic algorithm search, the PI controller parameter optimization speed is fast and robust, which solves the problem that the controller parameters are difficult to adapt to the changing environment, and improves the voltage detection accuracy and compensation performance of the series filter.

Claims

1. A method for detecting instantaneous voltage drop of a series filter, characterized in that: Specifically: (1) The distorted voltage value obtained by the voltage compensation instruction of the three-phase voltage based on the instantaneous reactive power dq0 method; (2) The set of distorted voltage values ​​is used as input values ​​to enter the PI controller. The PI controller optimizes the PI parameters based on the improved immune genetic algorithm. After the PWM signal is generated, it drives the inverter bridge to detect and compensate the voltage.

2. A method for detecting instantaneous voltage drop of a series filter according to claim 1, characterized in that: In step (1), the operation of obtaining the distorted voltage value is as follows: Assume that the three-phase voltage expression in the distribution network system is as shown in formula (1): In the formula, u a ,u b ,u c is the three-phase voltage of the distribution network, is the three-phase positive sequence component, as shown in formula (2): Where U + Refers to the maximum forward voltage, that is, the forward amplitude. Conversely, U - is the negative amplitude. After the nth harmonic is introduced, it is expressed as the positive and negative amplitude of the nth harmonic; ω refers to the angular frequency; ψ refers to the initial phase angle, which is the same as U and can be positive or negative. After the nth harmonic is introduced, it is expressed as the initial phase angle of the nth harmonic; t refers to the time variable, is the three-phase negative sequence component, as shown in formula (3): Where u0 is the zero-sequence component, as shown in equation (4): In the formula, is the positive sequence component of the nth harmonic, as shown in formula (5): In the formula, is the negative sequence component of the nth harmonic, as shown in equation (6): Three-phase voltage u dq0 and u abc The transformation relationship between them is obtained by the following formula (7): In the formula In the formula, The vector representing the positive sequence component Park; A vector representing the negative sequence component park; u 00 A vector representing the zero-sequence component. Filter out the high-order harmonics from equation (7) with a low-pass filter, and get the DC component expression (8): Performing dq0 inverse transformation on equation (8), we can obtain the fundamental voltage positive sequence component: As shown in formula (9): Then the instantaneous drop voltage detected by the series filter is as shown in formula (10):

3. A method for detecting instantaneous voltage drop of a series filter according to claim 2, characterized in that: In step (2), the mathematical model of the PI controller is as follows: Where K is the proportionality coefficient, K p is the voltage compensation parameter, is the integral time constant, e(t) is the deviation value of the PI controller input at time t; u(t) is the output value of the PI controller at time t; t is the time variable; in the actual digital control field, digital PI control is often used, using the summation form to replace the integral and the increment form to replace the differential. The approximate transformation is as follows: Where, T s is the sampling period, k is the sampling sequence, i is the variable parameter, and L represents the total number of sampling sequences. The discrete PI control expression can be obtained by simplifying the equation similar to e(kT) as shown in equation (13): In the above formula, u(k) is the output value at the kth sampling time, e(k) is the deviation value input at the kth sampling time, K I is the integration coefficient, u0 is the original initial value for starting the PI controller.

4. A method for detecting instantaneous voltage drop of a series filter according to claim 3, characterized in that In step (2), the specific principle of the PI controller parameter optimization method based on the improved immune genetic algorithm is: Will The parameter optimization sets an optimization objective function F, as shown in formula (14): Where T is the sampling time, T s is the number of sampling points, MinF represents the minimum value of the objective function F, e(t i ) is t i The deviation value input at the sampling time, i is the parameter variable, t i =iT S .

5. A method for detecting instantaneous voltage drop of a series filter according to claim 4, characterized in that: The steps of PI parameter optimization based on the improved immune genetic algorithm are as follows: (21) Select voltage compensation parameter K P , K I As the antigen Ag, using the forward constraint, N K I parameter variable, and K I (N+1) is set as antibody Ab; (22) Selection, crossover, mutation, and immunization; Calculate the fitness according to the PI performance optimization evaluation index of formula (14), and iterate using formula (13). If the iteration termination condition is met, the algorithm stops and outputs the optimal result, otherwise, repeat this step; (23) Generate the next generation of immune antibodies and determine whether the conditions are met; if so, stop and output the optimal result. In this way, the parameter K of the objective function takes the minimum value under the premise of satisfying the system stability. P (N+1), K I (N+1) is the global optimal solution K of the final PI parameter P , K I , otherwise return to step (2) and restart the fitness calculation.

6. A method for detecting instantaneous voltage drop of a series filter according to claim 5, characterized in that In step (22) and step (23), for each antigen Ag, the operation is as follows: (a) Option: Calculate each antibody Ab separately i and the same antigen Ag i The affinity coefficient β ij , directly call the shortest Euclidean distance formula to calculate: Where i represents the i-th generation, j represents the j-th, n represents the total number of antigens in the sample, and k = 1, 2, 3, ... n; (b) Crossover: According to β ij Screen and sort the antibodies, select the m antibodies with the highest affinity as network cells, and clone the m network cells; (c) Mutation: The cloned cells are subjected to mutation operation using the cell mutation formula C = C-α(CX), where C represents the cloned antibody cell, X represents the cloned antigen cell, and a represents the mutation rate; (d) Immunization: recalculate the affinity of C after the mutation operation, and select the one with the highest affinity C=60 as the partial memory cell dataset Mp; (e) Generate the next generation of new immune antibodies: Calculate the similarity S between each antibody ij , eliminate the similarity S in Mp ij Greater than the threshold σ s Individuals: (f) Merge Mp into the memory dataset M and update the dataset M.

7. A method for detecting instantaneous voltage drop of a series filter according to claim 5, characterized in that: In step (23), specifically: select from the N (0, 1) individuals generated in step (21), replace the individuals with poor affinity, and use the memory data set M obtained from the previous immune calculation as the antibody for the next generation of immune calculation, and return to step (22), calculate the fitness value of the population, and observe whether the termination condition is met.

Images