A method and system for pi optimization control based on an ac-ac converter

Abstract

The application discloses a PI optimization control method and system based on an AC-AC converter, and relates to the technical field of system closed-loop control. The method comprises the following steps: taking the control parameters of each loop PI controller in a data center power supply system as an individual, and constructing an initial population; determining the fitness of each individual in the initial population according to an optimization objective function; dividing the individuals in the initial population into three levels of roosters, hens and chicks; iteratively optimizing the roosters, hens and chicks through a position disturbance mechanism based on Cauchy distribution to obtain an optimal control parameter combination of the multiple loop PI controllers; and applying the optimal control parameter combination to the data center power supply system, so that the multiple loop PI controllers generate compensation instructions corresponding to different frequency harmonic components respectively, and drive the AC-AC converter to generate compensation current, thereby completing the PI control of the data center power supply system. The application realizes comprehensive optimization control of power quality of the data center power supply system.

Description

Technical Field

[0001] This invention relates to the field of closed-loop control technology, and in particular to a PI optimization control method and system based on an EQ converter. Background Technology

[0002] With increasing demands from electrical loads for power system quality, high reliability, energy saving, cost-effectiveness, and clean power supply have become the main development directions. High-frequency HVDC and UPS systems are widely used, improving energy efficiency through numerous frequency conversion energy-saving technologies. However, frequency conversion HVDC, UPS, inverters, and switching power supplies generate significant reactive power and harmonics. Direct AC-AC converters, based on traditional DC-DC converters, replace switching transistors and diodes with bidirectional switching transistors, enabling chopping of the input AC voltage. This new power quality control technology based on direct AC-AC conversion is a DC-AC conversion technology without a DC link. Because it avoids the low reliability and high cost problems caused by electrolytic capacitors, it possesses significant development potential and engineering application potential.

[0003] To simultaneously achieve harmonic suppression, scholars both domestically and internationally have proposed even-order harmonic modulation (EHM) methods. This involves adding a DC component and multiple even-order sinusoidal components to the duty cycle of the AC-AC converter, allowing the AC-AC converter to inject reactive power and odd-order harmonic compensation components into the power grid to improve the grid-side current waveform quality. This has led to a new type of active power filter without an inverter stage. However, research on AC-AC converter-based technology is still in its early stages, with significant room for further investigation into the coupling relationships between reactive power and harmonics, the coupling relationships between even-order harmonics, and decoupling strategies. When injecting even-order harmonics, each odd-order harmonic is influenced by multiple injected even-order harmonics. Coupling exists between the control of various even-order harmonics. When an AC-AC APF generates odd-order harmonic current compensation components at two adjacent frequencies, the EHM control variable corresponding to the higher-order compensation current component simultaneously affects the lower-order compensation current component—the EHM coupling problem. This complicates the precise control of each harmonic compensation current component and may even affect the system's stability.

[0004] Because of the coupling between the various control loops, parameter tuning is difficult and can easily cause system divergence. The existing control parameters use a trial-and-error method that relies on experience and has randomness, making it difficult to achieve good control results. Summary of the Invention

[0005] Therefore, it is necessary to provide a PI optimization control method and system based on an EQ converter to address the above-mentioned technical problems.

[0006] This invention provides a PI optimization control method based on an AC-AC converter, applied to a data center power supply system. The data center power supply system includes an AC-AC converter and multiple loop PI controllers for tracking different frequency harmonic components in the grid-side current. PI control methods include: The control parameters of each loop PI controller in the data center power supply system are treated as individuals to construct an initial population. The fitness of each individual in the initial population is determined based on the optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor. Based on the fitness of each individual, the individuals in the initial population are divided into three levels: roosters, hens, and chicks by improving the flocking algorithm. The improved flocking algorithm is obtained by introducing a position perturbation mechanism based on Cauchy distribution into the flocking algorithm. The rooster, hen, and chick are iteratively optimized using a Cauchy distribution-based positional perturbation mechanism to update the control parameters of each loop PI controller in the data center power supply system they represent. If the current iteration count exceeds the preset maximum iteration count, the iteration ends and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers. The optimal combination of control parameters is applied to the data center power supply system. Multiple loop PI controllers generate compensation commands corresponding to different frequency harmonic components, which drive the AC-AC converter to generate compensation current, thereby completing the PI control of the data center power supply system.

[0007] Optionally, the control parameters for each loop PI controller include: [ K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4 ]; in, K P1 , K I1 Let these be the proportional and integral constants of the PI controller in the first loop. K P2 , K I2These are the proportional and integral constants of the PI controller in the 5th loop. K P3 , K I3 For the proportional and integral constants of the PI controller in the 7th loop, K P4 , K I4 These are the proportional and integral constants of the PI controller in the 11th loop.

[0008] Optionally, the control parameters of each loop PI controller in the data center power supply system are used as individuals to construct an initial population, specifically including: An eight-dimensional vector is randomly generated based on the following formula as the initial individual, where the value of each dimension of the initial individual represents the control parameter of the PI controller in each loop of the data center power supply system: X i =[ K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4 ]; Based on the following formula, new individuals are generated according to the values ​​of each dimension of the initial individuals, resulting in a dataset: ; in, H i (-1, 1) and H i 0, i =1, 2, ...( N pop -1), j =1, 2..8, N pop Population size; The initial population is obtained by applying the following formula to the dataset: ; in, For the initial population, ub j As the upper bound of the dimension, lb jAs the lower bound of the dimension, For new individuals.

[0009] Optionally, the fitness of each individual in the initial population is determined based on the following formula: ; ; in, e 1 represents the tracking error of the fundamental frequency loop. e 2 represents the tracking error of the 5th loop. e 3 indicates the tracking error of the 7th loop. e 4 represents the tracking error of the 11th loop. t This represents the system runtime value. T min Indicates the initial time of system operation. T max Indicates the system's termination time. THD 1 represents the total harmonic distortion (THD) of the voltage waveform injected into the power grid by the AC / AC system. THD 2 represents the total harmonic distortion rate of the current waveform injected into the power grid by the AC-AC system. PF w1, w2, and w3 represent the power factor and weighting coefficients, respectively.

[0010] Optionally, based on the fitness of each individual, the individuals in the initial population are divided into three levels—roosters, hens, and chicks—by improving the flock algorithm, specifically including: The number of roosters is determined based on the following formula, and sorting indices from the first to the last are defined according to the number of roosters. RN The individual corresponding to this position is a rooster. RN =0.2* N pop ; The number of hens is determined based on the following formula, and a sorting index is defined according to the number of hens. RN +1 position to the RN + HN The individual corresponding to this position is a hen: HN =0.6* N pop ; The number of chicks is determined based on the following formula, and a sorting index is defined according to the number of chicks. RN + HN +1 position to the N pop The individual corresponding to this position is a chick: CN = N pop - RN - HN ; in, N pop For population size, RN The number of roosters. HN For the number of hens, CN This represents the number of chicks.

[0011] Optionally, the rooster can be iteratively optimized using a location perturbation mechanism based on Cauchy distribution, specifically including: For each individual rooster, the number of offspring is calculated based on the rooster's fitness using the following formula: ; in, For individual fitness, For maximum fitness, For maximum fitness, This represents the maximum number of offspring produced. This represents the minimum number of offspring produced. ; in, For the standard deviation of an individual, The maximum standard deviation of an individual. For the minimum standard deviation of an individual, The maximum number of iterations, k For the current iteration; The offspring locations of the rooster are generated using a location perturbation mechanism based on the Cauchy distribution, according to the following formula: ; Based on the ascending fitness order, as shown in the following formula, the position with the best fitness is selected as the rooster's position in the next iteration: ; ; ; in, The number of offspring of the rooster. For the rooster individual in the next iteration cycle.

[0012] Optionally, the hen can be iteratively optimized using a location perturbation mechanism based on Cauchy distribution, specifically including: For each hen, the hen's position is updated based on the following formula, taking into account the positional difference between the hen and its rooster and its competitive relationship with other hens: ; ; Where Rand1 and Rand2 are random numbers between 0 and 1, and r1 and r2 are natural numbers between 1 and N; The offspring of the hen are generated based on the following formula using a location perturbation mechanism based on Cauchy distribution: ; ; The position with the best fitness is selected as the position of the hen in the next iteration based on the following formula: ; ; ; in, For the mother hen, The number of offspring of the hen. For the new position of the hen's offspring.

[0013] Optionally, the chicks can be iteratively optimized using a location perturbation mechanism based on Cauchy distribution, specifically including: For each individual chick, the offspring position is generated using a Cauchy distribution-based positional perturbation mechanism based on the following formula: ; Based on the following formula, the top performers are selected according to fitness ranking. CN Each individual becomes a chick in the next iteration cycle: ; ; in, This represents the number of offspring of the chick.

[0014] This invention provides a PI optimization control system based on an EQ converter, comprising: The population construction module is used to construct an initial population by taking the control parameters of each loop PI controller in the data center power supply system as individuals. The fitness module is used to determine the fitness of each individual in the initial population based on the optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor. The partitioning module is used to divide individuals in the initial population into three levels: roosters, hens, and chicks, based on the fitness of each individual and through an improved flocking algorithm. The improved flocking algorithm is obtained by introducing a location perturbation mechanism based on Cauchy distribution into the flocking algorithm. The iterative module is used to iteratively optimize the rooster, hen, and chick through a positional perturbation mechanism based on Cauchy distribution, so as to update the control parameters of each loop PI controller in the data center power supply system they represent; if the current iteration number is greater than the preset maximum iteration number, the iteration ends and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers; The PI control module is used to apply the optimal combination of control parameters to the data center power supply system. It generates compensation commands corresponding to different frequency harmonic components through multiple loop PI controllers, drives the AC-AC converter to generate compensation current, and completes the PI control of the data center power supply system.

[0015] The PI optimization control method and system based on an inter-inter-interchange converter provided in this invention have the following advantages compared with the prior art: This invention transforms the complex PI parameter tuning process into an intelligent optimization problem by improving the chicken flock algorithm. By constructing a fitness function that integrates tracking error, harmonic distortion rate, and power factor, the algorithm is guided to automatically search for the optimal parameter combination, achieving fully automated parameter tuning and completely eliminating the reliance on manual trial and error experience.

[0016] More importantly, by incorporating the coupling effect between loops into the global fitness assessment through a location perturbation mechanism based on Cauchy distribution, decoupling control of each harmonic loop is indirectly achieved. This enables targeted output of compensation current, while simultaneously completing reactive power compensation and harmonic suppression, thereby achieving comprehensive optimization control of power quality in the data center power supply system. Attached Figure Description

[0017] Figure 1 Here is a main circuit diagram of a PI optimization control method based on an inter-inter-interchange converter provided in one embodiment; Figure 2 Here is a cross-cross converter APF topology diagram for a PI optimization control method based on a cross-cross converter provided in one embodiment; Figure 3 This is a control principle diagram of a PI optimization control method based on an inter-inter-interchange converter provided in one embodiment; Figure 4 This is a control flowchart of a PI optimization control method based on an EQ converter provided in one embodiment. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] This invention provides a PI optimization control method based on an AC-AC converter, applied to a data center power supply system. The data center power supply system includes an AC-AC converter and multiple loop PI controllers (Proportional-Integral Controllers) for tracking different frequency harmonic components in the grid-side current.

[0020] like Figure 4 As shown, the PI control method includes: The control parameters of each loop PI controller in the data center power supply system are treated as individuals to construct an initial population.

[0021] The fitness of each individual in the initial population is determined based on the optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor weighted.

[0022] Based on the fitness of each individual, the individuals in the initial population are divided into three levels: roosters, hens, and chicks by improving the flocking algorithm. The improved flocking algorithm is obtained by introducing a location perturbation mechanism based on Cauchy distribution into the flocking algorithm.

[0023] The rooster, hen, and chicks are iteratively optimized using a Cauchy distribution-based positional perturbation mechanism to update the control parameters of each loop PI controller in the data center power supply system they represent. If the current iteration count exceeds the preset maximum iteration count, the iteration ends, and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers.

[0024] The optimal combination of control parameters is applied to the data center power supply system. Multiple loop PI controllers generate compensation commands corresponding to different frequency harmonic components, which drive the AC-AC converter to generate compensation current, thereby completing the PI control of the data center power supply system.

[0025] A specific embodiment of the present invention is provided: I. Explanation of the principle.

[0026] In a three-phase three-wire balanced system, the current harmonic frequencies are predominantly 6kHz ± 1, mainly the 5th, 7th, 11th, and 13th harmonics. A direct AC-AC converter and even-harmonic modulation (EHM) technique are employed to generate corresponding capacitive odd-order harmonics, achieving reactive power compensation and harmonic suppression. The three-phase harmonic current detection method utilizes a multi-synchronous rotating coordinate transformation. For a specific frequency harmonic component in the grid-side current to be detected, a rotating coordinate transformation with the same angular velocity as the harmonic can convert the harmonic into a DC quantity. Then, a PI controller can be used for zero steady-state error tracking, employing multiple PI controls in the 1st, 5th, 7th, and 11th loops.

[0027] Control parameters: The control parameters for each loop PI controller are as follows: [ K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4 ]; in, K P1 , K I1 Let these be the proportional and integral constants of the PI controller in the first loop. K P2 , K I2 These are the proportional and integral constants of the PI controller in the 5th loop. K P3 , K I3 For the proportional and integral constants of the PI controller in the 7th loop, K P4 , K I4 These are the proportional and integral constants of the PI controller in the 11th loop.

[0028] The parameters output by the PI controller include: the EHM term coefficient in the duty cycle, which, after dq-abc coordinate transformation, becomes the 6k±2th order EHM term in the three-phase duty cycle, i.e.: d a ( t )= d a ( t ) 1th + d a ( t ) 5th + d a ( t ) 7th + d a ( t ) 11th ; d b ( t )=d b ( t ) 1th + d b ( t ) 5th + d b ( t ) 7th + d b ( t ) 11th ; d c ( t )= d c ( t ) 1th + d c ( t ) 5th + d c ( t ) 7th + d c ( t ) 11th ; constant term K Adding 0 to each EHM term yields the three-phase time-varying duty cycle. Then, modulating the duty cycle signal of each phase, the control signal of each switch in the entire three-phase AC-AC type APF system can be obtained.

[0029] II. Implementation of the Method.

[0030] In the PI control method, the control parameters are evaluated using the weighted superposition of the product of the square of the closed-loop tracking error of each loop and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor as the objective function for assessing control performance (Equation (3.1)). An improved flock optimization parameter solver is designed to optimize and tune the eight parameters of the PI controller. For example... Figure 3 As shown, the method includes the following steps:

[0031] 1. Using the Simulink module of MATLAB 2018b software, a simulation model of multiple PI controllers for a BUCK-type AC-AC converter was established, including AC power supply, reactive load, nonlinear load, insulated gate bipolar transistor (IGBT) with anti-parallel diode, filter capacitor, filter inductor, output resistor, power system simulation module Powergui, current and voltage measurement module, AD converter, filter capacitor, filter inductor, output resistor, reference value of output voltage, comparator module, control module, PWM modulation module, and IGBT drive module.

[0032] The transfer functions of the fundamental loop PI controller, the 5th loop PI controller, the 7th loop PI controller, and the 11th loop PI controller are as follows: ; ; ; ; 2. Configure the parameters of the parametric solver, including the population size. N pop Maximum number of iterations K max =200, Number of roosters RN =0.2* N pop Number of hens HN =0.6* N pop Number of chicks CN = N pop - RN - HN Initialize positive integers Itera-counter=0 and Itera-threshold=3, and reconstruct factors. G The initial value is G min Number of iterations k =0, G min =2, G max =20; 3. Based on FUCH chaos theory, generate the initial population. The individuals in the population POP are { X i}, i =1, 2, ... N pop An individual represents a decision variable, and a solution is a set of solutions, where the first individual represents a decision variable. i individual Xi This represents the eight control parameters for a fractional-order PID controller, namely... X i =[ K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4 The decision variables have 8 dimensions, and each dimension has its upper and lower bounds. ub j and lb j , j =1, 2..8. And i =1, 2, ... N pop , j =1, 2..8 refers to the first... k In the generation, the first i The first individual j The numerical value of the item dimension.

[0033] 3.1 Randomly generate an 8-dimensional vector, with each dimension... j All are within the range [0, 1] j =1, 2..8 are used as the initial individuals H 1，j .

[0034] 3.2 Substitute the value of each dimension of the initial individual into formula (1) to calculate and generate a new individual, which is the next individual.

[0035] (1) in, H i (-1, 1) and H i 0、 i =1, 2, ...( N pop -1) j =1, 2..8, N pop Population size.

[0036] 3.3 Repeat step 3.2 until the product is generated. Npop A new set of data was obtained by adding -1 new individuals. H i}, i =1, 2, ... N pop .

[0037] 3.4 Combine the data set generated above { H i}, i =1, 2, ... N pop Substituting into formula (2), calculate the initial population of the generation algorithm { }: (2) in, i =1, 2, ... N pop , j =1, 2..8, For the initial population, ub j As the upper bound of the dimension, lb j As the lower bound of the dimension, For new individuals.

[0038] 4. Set the current iteration number k For reconstruction factors G Perform a remainder operation to determine if the result is 0. If so, reorganize the population structure, randomly assign the corresponding individuals, increment the Itera-counter, and proceed to step 5. Otherwise, keep the current organizational structure unchanged and proceed to step 7.

[0039] 5. Set the initial population { }, i =1, 2, ... N pop Substitute the values ​​into the fitness function and calculate the fitness according to formula (3.1). f ( X i Meanwhile, the indexes are sorted in ascending order of fitness, as shown in formula (3.2), and the current optimal fitness and its corresponding position are selected.

[0040] (3.1) (3.2) in, e 1 represents the tracking error of the fundamental frequency loop. e 2 represents the tracking error of the 5th loop. e3 indicates the tracking error of the 7th loop. e 4 represents the tracking error of the 11th loop. t This represents the system runtime value. T min and T max These represent the system's initial and final times, respectively. THD 1 and THD 2 represents the total harmonic distortion (THD) of the voltage and current waveforms injected into the power grid by the AC / AC system, respectively. PF Let w1, w2, and w3 be the power factor, and w1, w2, and w3 be the weighting coefficients, satisfying w1 + w2 + w3 = 1.

[0041] 6. Based on formula (3.2), sort the index by fitness, establish the flock level, and define the first to the second position of the sorting index. RN The individual corresponding to this position is a rooster (the individual with the highest fitness rating): RN =0.2* N pop ; Define the sorted index. RN +1 position to the RN + HN The individual corresponding to this position is a hen (the individual with the middle fitness ranking): HN =0.6* N pop ; Define the sorted index. RN + HN +1 position to the N pop The individual corresponding to this position is a chick (the individual with the lowest fitness rank): CN = N pop - RN - HN ; in, N pop For population size, RN The number of roosters. HN For the number of hens, CN This represents the number of chicks.

[0042] Check if Itera-counter is less than Itera-threshold. If yes, proceed to the next step; otherwise, reset Itera-counter to zero. Determine the current... G Is it less than? G max If so, then G = G+1, if not, then G Remain unchanged.

[0043] 7. Substitute all individuals in the population into formula (4) and calculate the corresponding number of offspring according to fitness.

[0044] (4) in, For individual fitness, For maximum fitness, For maximum fitness, This represents the maximum number of offspring produced. This represents the minimum number of offspring produced.

[0045] (5) in, For the standard deviation of an individual, The maximum standard deviation of an individual. For the minimum standard deviation of an individual, The maximum number of iterations, k This is the current iteration.

[0046] 8. Among all individual roosters i =1, 2, ... RN Take one rooster at a time and repeat the following operation.

[0047] 8.1 The number of offspring calculated using formula (4) , i =1, 2, ... RN The variance is calculated using formula (5).

[0048] 8.2 Using formula (6), the offspring of this rooster were subjected to Cauchy distribution, and the results were obtained. The new position of each offspring.

[0049] (6) 8.3 Calculate the weight of the female rooster ( The fitness of the rooster and its newly generated offspring are sorted in ascending order of fitness as shown in formula (7). According to formulas (8) and (9), the first position in the sorted order is the position of the rooster in the next iteration cycle.

[0050] (7) (8) (9) in, The number of offspring of the rooster. For the individual rooster in the next iteration cycle, i =1, 2, ... RN .

[0051] 8.4 Continue using steps 8.1-8.3 until the desired result is obtained. RN A rooster , i =1, 2, ... RN The position of the next iteration cycle of}.

[0052] 9. In all individual hens i =1, 2, ... HN Take one hen at a time and repeat the following operations.

[0053] 9.1 Substitute into formula (10) to update the position of the hen.

[0054] (10) (11) (12) Where Rand1 and Rand2 are random numbers between 0 and 1; r1 is a natural number between 1 and N, which is the index number of the rooster; r2 is a natural number between 1 and N, which is the index number of the rooster or hen randomly selected from the population, and r1 is not equal to r2.

[0055] 9.2 The number of offspring of the hen after updating according to step 9.1, calculated using formula (4). , i =1, 2, ... HN The variance was calculated using formula (5). Using formula (6), the Cauchy distribution of the offspring produced by this hen was obtained. offspring The new position.

[0056] 9.3 Calculate the fitness of the hen and its offspring, and sort them in ascending order of fitness as shown in formula (13). According to formulas (14) and (15), the first rank is taken as the position of the hen in the next iteration cycle.

[0057] (13) (14) (15) in, For the mother hen, The number of offspring of the hen. For the new position of the hen's offspring.

[0058] 9.4 Continue using steps 9.1-9.3 until the desired result is obtained. HN A hen { , i =1, 2, ... HN The position of the next iteration cycle of}.

[0059] 10. Among all individual chicks i =1, 2, ... CN Take one chick at a time and repeat the following steps.

[0060] 10.1 The number of offspring of this chick calculated using formula (4) , i =1, 2, ... CN .

[0061] 10.2 Using formula (10), the offspring produced by this chick are subjected to chaotic perturbation to obtain... The new position of each offspring.

[0062] (16) (17) (18) in, i =1, 2, ... CN , j =1, 2... D .

[0063] Repeat the above steps until completion. D The update position of the chick is obtained according to formula (10).

[0064] (19) 10.3 Using step 10.2, let proceed in sequence A chaotic search was performed to obtain the chick's information. The position of each offspring.

[0065] 10.4 Continue with steps 10.2 and 10.3 until all steps are completed. CN Each chick produces offspring.

[0066] 10.5 Calculate the fitness of all female chicks and their offspring, and sort them in ascending order of fitness according to formula (20). According to formula (21), take the top... CN Each individual is a chick in the new iteration cycle.

[0067] (20) (twenty one) in, This represents the number of offspring of the chick.

[0068] 11. Determine the current iteration number k Is it less than the maximum number of iterations? K max So it is. k Increment by one and proceed to step 4; otherwise, output the position of the top-ranked rooster, which will be used as the optimal control parameter for the controller. K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4 ].

[0069] 12. Regarding the Simulink simulation model described in step 1 (such as...) Figure 1 The actual cross-connect converter corresponding to (as shown) Figure 2 As shown), the AC-AC converter hardware includes an AC power supply, an insulated gate bipolar transistor (IGBT) with an anti-parallel diode, a filter capacitor, a filter inductor, an output resistor, a current and voltage measurement module, an AD converter, a PWM modulation module, an IGBT drive module, an analog-to-digital converter, a digital signal processor (DSP), a current sensor, and a voltage sensor. Multiple PI controllers are implemented using the DSP, and the optimal control parameters of the PI controllers obtained in step 11 are... K P1 , K I1 , K P2 , K I2 , K P3 , K I3 , K P4 , K I4The input is fed into the three-phase Buck-type AC-AC converter control system to achieve optimized control of reactive power and harmonic suppression of the AC-AC converter. The power factor on the power supply side, the total harmonic distortion rate of the current and voltage on the bus are obtained by using an oscilloscope and a power quality analyzer.

[0070] Based on the same inventive concept, embodiments of the present invention also provide a PI optimization control system based on an EQ converter, the system comprising: The population construction module is used to construct an initial population by taking the control parameters of each loop PI controller in the data center power supply system as individuals.

[0071] The fitness module is used to determine the fitness of each individual in the initial population based on an optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor.

[0072] The partitioning module is used to divide individuals in the initial population into three levels—roosters, hens, and chicks—based on the fitness of each individual and through an improved flocking algorithm. The improved flocking algorithm is obtained by introducing a location perturbation mechanism based on Cauchy distribution into the flocking algorithm.

[0073] The iterative module is used to iteratively optimize the rooster, hen, and chick through a Cauchy distribution-based positional perturbation mechanism to update the control parameters of each loop PI controller in the data center power supply system they represent. If the current iteration count exceeds the preset maximum iteration count, the iteration ends, and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers.

[0074] The PI control module is used to apply the optimal combination of control parameters to the data center power supply system. It generates compensation commands corresponding to different frequency harmonic components through multiple loop PI controllers, drives the AC-AC converter to generate compensation current, and completes the PI control of the data center power supply system.

[0075] The embodiments described above are merely examples of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention.

Claims

1. A PI optimization control method based on an EQ converter, characterized in that, The data center power supply system includes an AC-AC converter and multiple loop PI controllers for tracking different frequency harmonic components in the grid-side current. The PI control method includes: The control parameters of each loop PI controller in the data center power supply system are treated as individuals to construct an initial population. The fitness of each individual in the initial population is determined based on the optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor. Based on the fitness of each individual, the individuals in the initial population are divided into three levels: roosters, hens, and chicks by an improved flocking algorithm. The improved flocking algorithm is obtained by introducing a position perturbation mechanism based on Cauchy distribution into the flocking algorithm. The rooster, hen, and chick are iteratively optimized using a Cauchy distribution-based positional perturbation mechanism to update the control parameters of each loop PI controller in the data center power supply system they represent. If the current iteration count exceeds the preset maximum iteration count, the iteration ends and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers. The optimal combination of control parameters is applied to the data center power supply system. Multiple loop PI controllers generate compensation commands corresponding to different frequency harmonic components, which drive the AC-AC converter to generate compensation current, thereby completing the PI control of the data center power supply system.

2. The PI optimization control method based on an inter-inter-interchange converter as described in claim 1, characterized in that, The control parameters for each loop PI controller include: [ K P1 ， K I1 ， K P2 ， K I2 ， K P3 ， K I3 ， K P4 ， K I4 ]； in, K P1 , K I1 Let these be the proportional and integral constants of the PI controller in the first loop. K P2 , K I2 These are the proportional and integral constants of the PI controller in the 5th loop. K P3 , K I3 For the proportional and integral constants of the PI controller in the 7th loop, K P4 , K I4 These are the proportional and integral constants of the PI controller in the 11th loop.

3. The PI optimization control method based on an inter-inter-interchange converter as described in claim 2, characterized in that, The step of constructing an initial population by treating the control parameters of each loop PI controller in the data center power supply system as individuals specifically includes: An eight-dimensional vector is randomly generated based on the following formula as the initial individual, where the value of each dimension of the initial individual represents the control parameter of each loop PI controller in the data center power supply system: X i ＝[ K P1 ， K I1 ， K P2 ， K I2 ， K P3 ， K I3 ， K P4 ， K I4 ]； Based on the following formula, new individuals are generated according to the values ​​of each dimension of the initial individual, resulting in a dataset: ； in, H i (-1, 1) and H i 0, i =1, 2, ...( N pop -1), j =1, 2..8, N pop Population size; The initial population is obtained by applying the following formula to the dataset: ； in, For the initial population, ub j As the upper bound of the dimension, lb j As the lower bound of the dimension, For new individuals.

4. The PI optimization control method based on an inter-inter-interchange converter as described in claim 1, characterized in that, The fitness of each individual in the initial population is determined based on the following formula: ； ； in, e 1 represents the tracking error of the fundamental frequency loop. e 2 represents the tracking error of the 5th loop. e 3 indicates the tracking error of the 7th loop. e 4 represents the tracking error of the 11th loop. t This represents the system runtime value. T min Indicates the initial time of system operation. T max Indicates the system's termination time. THD 1 represents the total harmonic distortion (THD) of the voltage waveform injected into the power grid by the AC / AC system. THD 2 represents the total harmonic distortion rate of the current waveform injected into the power grid by the AC-AC system. PF w1, w2, and w3 represent the power factor and weighting coefficients, respectively.

5. The PI optimization control method based on an inter-inter-interchange converter as described in claim 4, characterized in that, Based on the fitness of each individual, the improved flocking algorithm divides the individuals in the initial population into three levels: roosters, hens, and chicks. Specifically, this includes: The number of roosters is determined based on the following formula, and sorting indices from the first to the last are defined according to the number of roosters. RN The individual corresponding to this position is a rooster. RN =0.2* N pop ； The number of hens is determined based on the following formula, and a sorting index is defined according to the number of hens. RN +1 position to the RN + HN The individual corresponding to this position is a hen: HN =0.6* N pop ； The number of chicks is determined based on the following formula, and a sorting index is defined according to the number of chicks. RN + HN +1 position to the N pop The individual corresponding to this position is a chick: CN = N pop - RN - HN ； in, N pop For population size, RN The number of roosters. HN For the number of hens, CN This represents the number of chicks.

6. The PI optimization control method based on an inter-inter-interchange converter as described in claim 1, characterized in that, The iterative optimization of the rooster using a positional perturbation mechanism based on Cauchy distribution specifically includes: For each individual rooster, the number of offspring is calculated based on the rooster's fitness using the following formula: ； in, For individual fitness, For maximum fitness, For maximum fitness, This represents the maximum number of offspring produced. This represents the minimum number of offspring produced. ； in, For the standard deviation of an individual, The maximum standard deviation of an individual. For the minimum standard deviation of an individual, The maximum number of iterations, k For the current iteration; The offspring locations of the rooster are generated using a location perturbation mechanism based on the Cauchy distribution, according to the following formula: ； Based on the ascending fitness order, as shown in the following formula, the position with the best fitness is selected as the rooster's position in the next iteration: ； ； ； in, The number of offspring of the rooster. For the rooster individual in the next iteration cycle.

7. The PI optimization control method based on an inter-inter-interchange converter as described in claim 1, characterized in that, The iterative optimization of the hen through a location perturbation mechanism based on Cauchy distribution specifically includes: For each hen, the hen's position is updated based on the following formula, taking into account the positional difference between the hen and its rooster and its competitive relationship with other hens: ； ； Where Rand1 and Rand2 are random numbers between 0 and 1, and r1 and r2 are natural numbers between 1 and N; The offspring of the hen are generated based on the following formula using a location perturbation mechanism based on Cauchy distribution: ； ； The position with the best fitness is selected as the position of the hen in the next iteration based on the following formula: ； ； ； in, For the mother hen, The number of offspring of the hen. For the new position of the hen's offspring.

8. The PI optimization control method based on an inter-inter-interchange converter as described in claim 1, characterized in that, The iterative optimization of the chicks using a location perturbation mechanism based on Cauchy distribution specifically includes: For each individual chick, the offspring position is generated using a Cauchy distribution-based positional perturbation mechanism based on the following formula: ； Based on the following formula, the top performers are selected according to fitness ranking. CN Each individual becomes a chick in the next iteration cycle: ； ； in, This represents the number of offspring of the chick.

9. A PI optimization control system based on an EQ converter, characterized in that, include: The population construction module is used to construct an initial population by taking the control parameters of each loop PI controller in the data center power supply system as individuals. The fitness module is used to determine the fitness of each individual in the initial population based on an optimization objective function composed of the product of the square of the closed-loop tracking error of each loop in the data center power supply system and time, the total harmonic distortion rate of the system output voltage waveform, and the power factor. The partitioning module is used to divide individuals in the initial population into three levels: roosters, hens, and chicks, based on the fitness of each individual and through an improved flocking algorithm. The improved flocking algorithm is obtained by introducing a positional perturbation mechanism based on Cauchy distribution into the flocking algorithm. The iterative module is used to iteratively optimize the rooster, hen, and chick through a Cauchy distribution-based positional perturbation mechanism to update the control parameters of the PI controllers of each loop in the data center power supply system they represent. If the current iteration count is greater than the preset maximum iteration count, the iteration ends and the individual with the highest fitness is taken as the optimal combination of control parameters for multiple loop PI controllers; The PI control module is used to apply the optimal combination of control parameters to the data center power supply system. It generates compensation commands corresponding to different frequency harmonic components through multiple loop PI controllers, drives the AC-AC converter to generate compensation current, and completes the PI control of the data center power supply system.

Images