Photovoltaic inverter parameter identification method based on improved subtraction average optimizer algorithm

Abstract

The invention discloses a photovoltaic inverter parameter identification method based on an improved subtraction averaging optimizer algorithm, and the method comprises the steps: building an identification model based on a photovoltaic hardware-in-the-loop test platform according to an existing photovoltaic hardware-in-the-loop test model, and carrying out the identification of the parameters of a photovoltaic inverter based on the photovoltaic hardware-in-the-loop test data. Identifying the control parameters of the photovoltaic inverter in the identification model by using an improved subtraction average optimizer algorithm; according to the method, a golden sine and chaotic mapping strategy is introduced on the basis of a subtraction average optimizer algorithm, the algorithm is effectively prevented from falling into local optimum in the photovoltaic inverter parameter identification process through the excellent global search capability of the golden sine strategy, the randomness and diversity of particles can be enhanced through the chaotic mapping strategy, and the photovoltaic inverter parameter identification accuracy is improved. According to the improved algorithm, the distribution of the particles in the search space is more uniform, so that the identification efficiency and precision in the parameter identification process of the photovoltaic inverter are effectively improved by adopting the improved algorithm.

Description

Technical Field

[0001] The invention belongs to the technical field of photovoltaic grid-connected power generation, and in particular relates to a photovoltaic inverter parameter identification method based on an improved subtraction average optimizer algorithm. Background Art

[0002] Accurate inverter control parameters play a vital role in the dynamic stability analysis of photovoltaic systems. As a key component in photovoltaic systems, the inverter is responsible for converting the DC power of the photovoltaic array into AC power and supplying it to the grid or load. The accuracy of the control parameters in its model directly affects the stability of the entire photovoltaic system. However, due to the manufacturer's principle of protecting trade secrets, it is difficult to obtain accurate model parameters to establish an equivalent model consistent with the actual photovoltaic unit, and thus it is impossible to accurately simulate the behavioral characteristics of the photovoltaic unit in actual operation and conduct in-depth dynamic stability analysis. In order to overcome this problem, measured data is usually used to identify the control parameters to obtain accurate and reliable parameters to establish a model that matches the output characteristics of the actual photovoltaic unit.

[0003] At present, the methods of photovoltaic inverter parameter identification can be roughly divided into two categories: traditional identification algorithms and intelligent identification algorithms. Traditional identification algorithms, represented by least squares method and extended Kalman filter, are intuitive and easy to implement, and are the cornerstone of parameter identification. However, when faced with complex multi-parameter identification problems, traditional algorithms often face the bottleneck of improving identification accuracy and it is difficult to achieve high-precision identification effects. In contrast, intelligent identification algorithms can flexibly respond to the challenges of multi-parameter identification and achieve synchronous and accurate identification of multiple parameters. Intelligent identification algorithms not only improve the accuracy and efficiency of identification, but also broaden the application scope of parameter identification, providing strong technical support for the optimal design and efficient operation and maintenance of photovoltaic systems; therefore, it is necessary to propose a photovoltaic inverter parameter identification method based on an improved subtraction average optimizer algorithm to solve the above problems. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a photovoltaic inverter parameter identification method based on an improved subtraction average optimizer algorithm, aiming to solve the problem that the prior art adopts the subtraction average optimizer algorithm and easily falls into the local optimum, resulting in inaccurate results. By introducing the golden sine and chaos mapping strategies on the basis of the subtraction average optimizer algorithm, the identification efficiency and accuracy in the photovoltaic inverter parameter identification process are effectively improved.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is: The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm includes: S1, based on the photovoltaic hardware-in-the-loop test platform, builds an identification model according to the existing photovoltaic hardware-in-the-loop test model, including: S101, based on the photovoltaic hardware-in-the-loop test platform, collecting photovoltaic inverter output response data under multiple working conditions as identification data of the identification model; S102, perform a photovoltaic hardware-in-the-loop test, determine the control method adopted by the photovoltaic controller hardware-in-the-loop according to the photovoltaic hardware-in-the-loop test data, build a photovoltaic inverter control model based on this, and determine the parameters to be identified; S2, based on the photovoltaic hardware-in-the-loop test data, uses the improved subtraction average optimizer algorithm to identify the photovoltaic inverter control parameters in the identification model.

[0006] Preferably, in step S1, the DC bus capacitor and the AC filter capacitor and inductor in the identification model are consistent with the hardware-in-the-loop.

[0007] Preferably, step S102 of establishing a control model of a photovoltaic inverter includes the following process: The inverter adopts a dual-loop control strategy of voltage outer loop and current inner loop. In steady-state operation, the active power output by the grid-connected inverter is controlled by the voltage outer loop, and the grid-side power factor is controlled by the current inner loop. The q-axis current reference value is set to 0 to ensure that the inverter operates at a unity power factor. The specific mathematical model of the voltage outer loop controller is: ; In the formula, u dc, u dc* are the actual and reference voltages of the controller DC side respectively; Q , Q *Respectively actual and reference reactive power of the controller; k PV, k IV are the proportional and integral coefficients of the voltage outer loop control respectively; k PQ, k IQ are the proportional and integral coefficients of the reactive outer loop control respectively; i gd* 、i gq* are the reference values ​​of the d-axis and q-axis components of the inner current loop respectively; S is the Latent transformation differential operator; Under fault conditions, the active and reactive currents are determined by the low-through control module. The mathematical model during the low-through period is shown below: ; In the formula, and They are respectively the active and reactive support current during low-voltage wear-through period, k 1_Id, k 2_Id and idset are active current calculation coefficient 1, active current calculation coefficient 2 and active current calculation coefficient 3 respectively; k 1_Iq, k 2_Iq and i qset are reactive current calculation coefficient 1, reactive current calculation coefficient 2 and reactive current calculation coefficient 3 respectively; v Lin is the low voltage ride-through threshold; v t is the terminal voltage amplitude; after the fault is cleared, the voltage ride-through recovery process begins, and the active current follows the specified slope K recover; The controller model of the current inner loop is: ; In the formula, k PI1, k II1 are the proportional and integral parameters of the inner loop of the d-axis current respectively; k PI2, k II2 are the proportional and integral parameters of the inner loop of the q-axis current respectively; u sd and u sq are the d-axis and q-axis voltage components of the photovoltaic power port respectively; i gd 、i gq are the actual d-axis and q-axis components of the inner loop current respectively; L is the equivalent inductance between the inverter outlet and the grid connection point; ω 1 is the synchronous angular velocity; Based on the above photovoltaic inverter control model, the control parameters include k PV, k IV. k PQ, k IQ, k PI1, k II1, k PI2, k II2, k 1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K ; In the actual model, the proportional and integral coefficients of the voltage outer loop are equal to each other, and the proportional and integral coefficients of the current inner loop are equal to each other, that is: ; The parameters to be identified in step S1 include k PV, k IV. k PI1, k II1, k1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K . Preferably, step S2 comprises: S201, using Logistic chaotic mapping to randomly generate individuals of the initial population in the exploration space, the Logistic mapping is as follows: ; In the formula, r is the chaos control parameter, when r ∈[3.57,4], the mapping exhibits chaotic behavior; x n is the value of the current solution, x Initial value of n x 0 is randomly generated in (0,1); for each individual i =1,2,…,N generates a chaotic sequence of length D. The chaotic sequence is recursively generated according to the above formula x 1, x 2, ..., x D; The chaotic sequence x n∈(0,1) is mapped to the upper and lower bounds of the solution as follows: ; In the formula, X n is the initial population generated, lb is the lower bound of the solution space, ub is the upper bound of the solution space; S202, calculate fitness, after individual initialization, assign values ​​to each parameter of the particle, run the identification model and obtain the output active current i gd_a and reactive current i gq_a, and compare it with the active current in the measured data i gd_e and reactive current i The current error of gq_e is used as the fitness of the particle and calculated: ; In the formula, J is the fitness function value; n is the total length of the data; through the fitness function, the optimal value of the individual particle is obtained and recorded; S203, integrating golden sine adjustment step length: Furthermore, the golden sine is a sine function based on the golden ratio, which can produce a special regular step size change during the optimization process, thereby balancing the needs of global search and local convergence; the golden ratio It is approximately 1.618, which has a unique regularity in optimization. By combining the golden ratio with the sine function, the step size of the optimization algorithm can be dynamically adjusted; The basic form of the golden sine function is as follows: ; In the formula, t is the number of iterations, and the value generated by sin fluctuates in the range of [-1,1]. In the subtraction mean optimization algorithm, the golden sine is introduced to dynamically adjust the step size factor: ; In the formula, α ( t ) is the adaptive step size factor in the subtraction operation; β ( t ) is the adaptive step size factor in the averaging operation; α max, α min are the maximum and minimum values ​​of the step factor in the subtraction operation; β max, β min are the maximum and minimum values ​​of the step factor in the averaging operation, respectively; S204, update individual position: The core steps of the subtraction average optimizer algorithm are used to update the individual positions—subtraction and averaging operations; at each iteration, the individual positions are updated according to the current optimal solution. X Best and population average positions Update individual locations; Subtraction operation: In each iteration, the position of the individual is updated by a subtraction operation: ; Average operation: The updated individuals are further adjusted in position through the average operation: ; The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm takes the average error of active current and reactive current as the evaluation index to verify the effectiveness of the photovoltaic inverter identification result. The error calculation formula is as follows: ; In the formula, F Id and F Iq is the average error of active current and reactive current, K e_start, and K e_end are the first and last data serial numbers respectively; Ida and Iqa are the active current and reactive current of the measured data respectively; Ide and Iqe are the active current and reactive current of the identified data respectively.

[0008] The beneficial effects of the present invention are as follows: 1. This method introduces the golden sine and chaos mapping strategies on the basis of the subtraction average optimizer algorithm. The subtraction average optimizer algorithm has a simple principle and fast optimization speed, but when identifying multiple control parameters of the photovoltaic system, it may fall into the local optimum, resulting in inaccurate identification results. The golden sine strategy has excellent global search ability, which effectively avoids the algorithm from falling into the "local optimum" during the photovoltaic inverter parameter identification process. The chaos mapping strategy can enhance the randomness and diversity of particles and make the distribution of particles in the search space more uniform. Therefore, the improved algorithm is adopted to effectively improve the identification efficiency and accuracy in the photovoltaic inverter parameter identification process.

[0009] 2. The method of the present invention uses Logistic chaotic mapping and golden sine function to improve the initial value and step size in the subtraction average optimizer algorithm, accelerates the convergence speed, and effectively solves the problem of falling into the local optimal solution in the search space.

[0010] 3. The present invention identifies the control parameters of the photovoltaic system model through an improved subtraction average optimizer algorithm, which can effectively improve the parameter identification efficiency and accuracy, and enhance the control performance and adaptability of complex nonlinear systems.

[0011] 4. The method of the present invention identifies the parameters of the photovoltaic inverter black box controller based on the measured data of hardware in the loop. The control parameters during steady state and transient state are considered in the identification process, which can effectively reflect the actual operating characteristics of the photovoltaic inverter and accurately analyze its impact on the actual system operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 It is a schematic diagram of the flow chart of the present invention; Figure 2 A control block diagram of a photovoltaic controller model in an embodiment of the present invention; Figure 3 This is a comparison diagram of d-axis current identification data and measured data of the photovoltaic system identification model in the embodiment of the present invention under working condition 1; Figure 4 This is a comparison diagram of q-axis current identification data and measured data of the photovoltaic system identification model in the embodiment of the present invention under working condition 1; Figure 5 This is a comparison diagram of d-axis current identification data and measured data of the photovoltaic system identification model in the embodiment of the present invention under working condition 2; Figure 6 This is a comparison chart of the q-axis current identification data and the measured data of the photovoltaic system identification model under working condition 2 in an embodiment of the present invention. DETAILED DESCRIPTION

[0013] Embodiment 1: like Figure 1 As shown, a photovoltaic inverter parameter identification method based on an improved subtraction average optimizer algorithm includes: S1, based on the photovoltaic hardware-in-the-loop test platform, builds an identification model according to the existing photovoltaic hardware-in-the-loop test model, including: S101, based on the photovoltaic hardware-in-the-loop test platform, collecting photovoltaic inverter output response data under multiple working conditions as identification data of the identification model; S102, perform a photovoltaic hardware-in-the-loop test, determine the control method adopted by the photovoltaic controller hardware-in-the-loop according to the photovoltaic hardware-in-the-loop test data, build a photovoltaic inverter control model based on this, and determine the parameters to be identified; S2, based on the photovoltaic hardware-in-the-loop test data, uses the improved subtraction average optimizer algorithm to identify the photovoltaic inverter control parameters in the identification model.

[0014] Preferably, in step S1, the DC bus capacitor and the AC filter capacitor and inductor in the identification model are consistent with the hardware-in-the-loop.

[0015] Embodiment 2: like Figure 2 As shown, step S102 of establishing a control model of a photovoltaic inverter includes the following process: The inverter adopts a dual-loop control strategy of voltage outer loop and current inner loop. In steady-state operation, the active power output by the grid-connected inverter is controlled by the voltage outer loop, and the grid-side power factor is controlled by the current inner loop. The q-axis current reference value is set to 0 to ensure that the inverter operates at a unity power factor. The specific mathematical model of the voltage outer loop controller is: ; In the formula, u dc, u dc* are the actual and reference voltages of the controller DC side respectively; Q , Q *Respectively actual and reference reactive power of the controller; k PV, k IV are the proportional and integral coefficients of the voltage outer loop control respectively; k PQ, k IQ are the proportional and integral coefficients of the reactive outer loop control respectively; i gd* 、i gq* are the reference values ​​of the d-axis and q-axis components of the inner current loop respectively; S is the Latent transformation differential operator; Under fault conditions, the active and reactive currents are determined by the low-through control module. The mathematical model during the low-through period is shown below: ; In the formula, and They are respectively the active and reactive support current during low-voltage wear-through period, k 1_Id, k 2_Id and i dset are active current calculation coefficient 1, active current calculation coefficient 2 and active current calculation coefficient 3 respectively; k 1_Iq, k 2_Iq and i qset are reactive current calculation coefficient 1, reactive current calculation coefficient 2 and reactive current calculation coefficient 3 respectively; v Lin is the low voltage ride-through threshold; v t is the terminal voltage amplitude; after the fault is cleared, the voltage ride-through recovery process begins, and the active current follows the specified slope K recover; The controller model of the current inner loop is: ; In the formula, k PI1, k II1 are the proportional and integral parameters of the inner loop of the d-axis current respectively; k PI2, k II2 are the proportional and integral parameters of the inner loop of the q-axis current respectively; u sd and u sq are the d-axis and q-axis voltage components of the photovoltaic power port respectively; i gd 、i gq are the actual d-axis and q-axis components of the inner loop current respectively; L is the equivalent inductance between the inverter outlet and the grid connection point; ω 1 is the synchronous angular velocity; Based on the above photovoltaic inverter control model, the control parameters include k PV, k IV. k PQ, k IQ, k PI1, k II1, k PI2, k II2, k 1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K ; In the actual model, the proportional and integral coefficients of the voltage outer loop are equal to each other, and the proportional and integral coefficients of the current inner loop are equal to each other, that is: ; The parameters to be identified in step S1 include k PV, k IV. k PI1, k II1, k 1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K . Preferably, step S2 comprises: S201, using Logistic chaotic mapping to randomly generate individuals of the initial population in the exploration space, the Logistic mapping is as follows: ; In the formula, r is the chaos control parameter, when r ∈[3.57,4], the mapping exhibits chaotic behavior; x n is the value of the current solution, x Initial value of n x 0 is randomly generated in (0,1); for each individual i =1,2,…,N generates a chaotic sequence of length D. The chaotic sequence is recursively generated according to the above formula x 1, x 2, ..., x D; The chaotic sequence x n∈(0,1) is mapped to the upper and lower bounds of the solution as follows: ; In the formula, X n is the initial population generated, lb is the lower bound of the solution space, ub is the upper bound of the solution space; S202, calculate fitness, after individual initialization, assign values ​​to each parameter of the particle, run the identification model and obtain the output active current i gd_a and reactive current i gq_a, and compare it with the active current in the measured data i gd_e and reactive current i The current error of gq_e is used as the fitness of the particle and calculated: ; In the formula, Jis the fitness function value; n is the total length of the data; through the fitness function, the optimal value of the individual particle is obtained and recorded; S203, integrating golden sine adjustment step length: Furthermore, the golden sine is a sine function based on the golden ratio, which can produce a special regular step size change during the optimization process, thereby balancing the needs of global search and local convergence; the golden ratio It is approximately 1.618, which has a unique regularity in optimization. By combining the golden ratio with the sine function, the step size of the optimization algorithm can be dynamically adjusted; The basic form of the golden sine function is as follows: ; In the formula, t is the number of iterations, and the value generated by sin fluctuates in the range of [-1,1]. In the subtraction mean optimization algorithm, the golden sine is introduced to dynamically adjust the step size factor: ; In the formula, α ( t ) is the adaptive step size factor in the subtraction operation; β ( t ) is the adaptive step size factor in the averaging operation; α max, α min are the maximum and minimum values ​​of the step factor in the subtraction operation; β max, β min are the maximum and minimum values ​​of the step factor in the averaging operation, respectively; S204, update individual position: The core steps of the subtraction average optimizer algorithm are used to update the individual positions—subtraction and averaging operations; at each iteration, the individual positions are updated according to the current optimal solution. X Best and population average positions Update individual locations; Subtraction operation: In each iteration, the position of the individual is updated by a subtraction operation: ; Average operation: The updated individuals are further adjusted in position through the average operation: ; The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm takes the average error of active current and reactive current as the evaluation index to verify the effectiveness of the photovoltaic inverter identification result. The error calculation formula is as follows: ; In the formula, FId and F Iq is the average error of active current and reactive current, K e_start, and K e_end are the first and last data serial numbers respectively; Ida and Iqa are the active current and reactive current of the measured data respectively; Ide and Iqe are the active current and reactive current of the identified data respectively.

[0016] Embodiment three: Such as 3 and Figure 4 As shown, this embodiment discloses the identification and actual data response comparison of active current Id and reactive current Iq when the inverter outlet voltage drops to 50% of the rated voltage; Figure 3 The dynamic response of the Id component is shown, reflecting the rapid recovery of the system after being disturbed. The identification and measured data are almost completely consistent, indicating that the parameter identification model can accurately fit the dynamic behavior of the actual system. Figure 4 It shows that the Iq component reflects the transient characteristics of reactive current during disturbance. The identification data also matches the measured data well, indicating that the identification model has a strong ability to track the dynamic changes of reactive current.

[0017] Figure 5 and Figure 6 The following are the comparison diagrams of the identification and actual data responses of the active current Id and reactive current Iq when the inverter outlet voltage drops to 35% of the rated voltage. The identification and measured data are almost completely consistent, indicating that the identification model has excellent fitting effect on the dynamic characteristics of the system and has good adaptability to various working conditions. Figures 3 to 6 Perform error calculation: ; The calculated maximum deviation of active current between the identification data and the measured data does not exceed 0.0018, and the maximum deviation of reactive current does not exceed 0.0039, which confirms the high accuracy of the identification model and the effectiveness of the identification method.

Claims

1. A photovoltaic inverter parameter identification method based on an improved subtraction average optimizer algorithm, characterized in that: include: S1, based on the photovoltaic hardware-in-the-loop test platform, builds an identification model according to the existing photovoltaic hardware-in-the-loop test model, including: S101, based on the photovoltaic hardware-in-the-loop test platform, collecting photovoltaic inverter output response data under multiple working conditions as identification data of the identification model; S102, perform a photovoltaic hardware-in-the-loop test, determine the control method adopted by the photovoltaic controller hardware-in-the-loop according to the photovoltaic hardware-in-the-loop test data, build a photovoltaic inverter control model based on this, and determine the parameters to be identified; S2, based on the photovoltaic hardware-in-the-loop test data, uses the improved subtraction average optimizer algorithm to identify the photovoltaic inverter control parameters in the identification model.

2. The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm according to claim 1 is characterized in that: In the step S1, the DC bus capacitor and the AC filter capacitor and inductor in the identification model are consistent with the hardware-in-the-loop.

3. The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm according to claim 1 is characterized in that: Step S102 establishes a control model for the photovoltaic inverter, including the following process: The inverter adopts a dual-loop control strategy of voltage outer loop and current inner loop. In steady-state operation, the active power output by the grid-connected inverter is controlled by the voltage outer loop, and the grid-side power factor is controlled by the current inner loop. The q-axis current reference value is set to 0 to ensure that the inverter operates at a unity power factor. The specific mathematical model of the voltage outer loop controller is: ； In the formula, u dc, u dc* are the actual and reference voltages of the controller DC side respectively; Q , Q *Respectively actual and reference reactive power of the controller; k PV, k IV are the proportional and integral coefficients of the voltage outer loop control respectively; k PQ, k IQ are the proportional and integral coefficients of the reactive outer loop control respectively; i gd* 、i gq* are the reference values ​​of the d-axis and q-axis components of the inner current loop respectively; S is the Latent transformation differential operator; Under fault conditions, the active and reactive currents are determined by the low-through control module. The mathematical model during the low-through period is shown below: ； In the formula, and They are respectively the active and reactive support current during low-voltage wear-through period, k 1_Id, k 2_Id and i dset are active current calculation coefficient 1, active current calculation coefficient 2 and active current calculation coefficient 3 respectively; k 1_Iq, k 2_Iq and i qset are reactive current calculation coefficient 1, reactive current calculation coefficient 2 and reactive current calculation coefficient 3 respectively; v Lin is the low voltage ride-through threshold; v t is the terminal voltage amplitude; after the fault is cleared, the voltage ride-through recovery process begins, and the active current follows the specified slope K recover; The controller model of the current inner loop is: ； In the formula, k PI1, k II1 are the proportional and integral parameters of the inner loop of the d-axis current respectively; k PI2, k II2 are the proportional and integral parameters of the inner loop of the q-axis current respectively; u sd and u sq are the d-axis and q-axis voltage components of the photovoltaic power port respectively; i gd 、i gq are the actual d-axis and q-axis components of the inner loop current respectively; L is the equivalent inductance between the inverter outlet and the grid connection point; ω 1 is the synchronous angular velocity; Based on the above photovoltaic inverter control model, the control parameters include k PV, k IV. k PQ, k IQ, k PI1, k II1, k PI2, k II2, k 1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K ; In the actual model, the proportional and integral coefficients of the voltage outer loop are equal to each other, and the proportional and integral coefficients of the current inner loop are equal to each other, that is: ； The parameters to be identified in step S1 include k PV, k IV. k PI1, k II1, k 1_Id, k 2_Id, i dset, k 1_Iq, k 2_Iq, i qset and K .

4. The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm according to claim 3 is characterized in that: Step S2 includes: S201, using Logistic chaotic mapping to randomly generate individuals of the initial population in the exploration space, the Logistic mapping is as follows: ； In the formula, r is the chaos control parameter, when r ∈[3.57,4], the mapping exhibits chaotic behavior; x n is the value of the current solution, x Initial value of n x 0 is randomly generated in (0,1); for each individual i =1,2,…,N generates a chaotic sequence of length D. The chaotic sequence is recursively generated according to the above formula x 1, x 2, ..., x D; The chaotic sequence x n∈(0,1) is mapped to the upper and lower bounds of the solution as follows: ； In the formula, X n is the initial population generated, lb is the lower bound of the solution space, ub is the upper bound of the solution space; S202, calculate fitness, after individual initialization, assign values ​​to each parameter of the particle, run the identification model and obtain the output active current i gd_a and reactive current i gq_a, and compare it with the active current in the measured data i gd_e and reactive current i The current error of gq_e is used as the fitness of the particle and calculated: ； In the formula, J is the fitness function value; n is the total length of the data; through the fitness function, the optimal value of the individual particle is obtained and recorded; S203, the golden sine is integrated to adjust the step size. The basic form of the golden sine function is as follows: ； In the formula, t is the number of iterations, and the value generated by sin fluctuates in the range of [-1,1]. In the subtraction mean optimization algorithm, the golden sine is introduced to dynamically adjust the step size factor: ； In the formula, α ( t ) is the adaptive step size factor in the subtraction operation; β ( t ) is the adaptive step size factor in the averaging operation; α max, α min are the maximum and minimum values ​​of the step factor in the subtraction operation; β max, β min are the maximum and minimum values ​​of the step factor in the averaging operation, respectively; S204, update individual position: The core steps of the subtraction average optimizer algorithm are used to update the individual positions—subtraction and averaging operations; at each iteration, the individual positions are updated according to the current optimal solution. X Best and population average positions Update individual locations; Subtraction operation: In each iteration, the position of the individual is updated by a subtraction operation: ； Average operation: The updated individuals are further adjusted in position through the average operation: ； The photovoltaic inverter parameter identification method based on the improved subtraction average optimizer algorithm takes the average error of active current and reactive current as the evaluation index to verify the effectiveness of the photovoltaic inverter identification result. The error calculation formula is as follows: ； In the formula, F Id and F Iq is the average error of active current and reactive current, K e_start, and K e_end are the first and last data serial numbers respectively; Ida and Iqa are the active current and reactive current of the measured data respectively; Ide and Iqe are the active current and reactive current of the identified data respectively.

Images