A dual-closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer

Abstract

This invention discloses a dual-closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer, belonging to the field of three-phase photovoltaic grid-connected inverters. The steps are as follows: establishing a mathematical model of the inverter, determining i... d It is the stable DC bus voltage U dc The control signal; a new LADRC controller is established based on a two-stage coupled extended state observer; a high-performance dual-closed-loop grid-connected control system is constructed based on the LADRC controller and the QPR controller. The LADRC controller acting on the voltage outer loop is equipped with a two-stage coupled extended state observer, which can achieve more efficient suppression and compensation for interference components in different frequency bands. In conjunction with the QPR controller acting on the current inner loop, the robustness and stability of the grid-connected system controller are improved.

Description

Technical Field

[0001] This invention belongs to the field of three-phase photovoltaic grid-connected inverter technology, specifically relating to a dual closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer. Background Technology

[0002] With the rapid expansion of renewable energy generation, three-phase grid-connected inverters have become core equipment for efficiently converting DC power sources such as solar and wind power and connecting them to the grid. During operation, the inverter's DC bus voltage must remain stable to ensure continuous bidirectional energy flow and high-quality grid-side current output. However, due to uncertainties such as load fluctuations, grid fluctuations, and the characteristics of power devices, the DC bus voltage may fluctuate significantly in both dynamic and steady-state processes. This not only affects the inverter's reliability and efficiency but may also lead to power quality degradation or trigger protection actions. Furthermore, the increasing diversification and complexity of power systems makes maintaining DC bus voltage stability a critical bottleneck.

[0003] In the control process of grid-connected inverters, energy balance mainly relies on DC-side voltage regulation control, which typically exhibits a slow dynamic response; while current-limiting control for maximizing the power factor requires a faster response speed. Therefore, the difference in control performance between the two places higher demands on the design of control strategies.

[0004] In the dual-loop control strategy, the outer loop regulates the DC bus energy balance, mapping the bus voltage error to a d-axis active current reference; the inner loop implements a current closed loop within the grid synchronous coordinate system, ensuring that the modulation waveform follows the command in real time. The two loops cooperate to achieve the control objective of "bus voltage stabilization + high grid-connected power factor". Currently, commonly used controllers for grid-connected inverters include traditional proportional-integral (PI) controllers, proportional-integral-derivative (PID) controllers, and model-based predictive control (MPC), sliding mode control (SMC), and adaptive control methods. PI and PID controllers are simple in structure and easy to implement, but their robustness to system parameter changes and external disturbances is limited; while advanced control methods such as MPC and SMC have strong robustness and disturbance rejection capabilities, they require high accuracy of the system model and complexity of the control algorithm, limiting their application to variations in actual operating conditions and model uncertainties. In reality, grid-connected systems are often complex and nonlinear. Coupled with various disturbances, it is difficult to establish accurate mathematical models of these systems. Therefore, voltage outer-loop controllers or compensation methods based on specific mathematical models will distort the control results as the control process continues, leading to a decrease in the grid power factor. While Active Disturbance Rejection Control (ADRC) can quickly estimate internal and external disturbances, such as incomplete system models, into a total disturbance using an observer, and cancel them out in the linear error feedback rate, achieving real-time disturbance rejection, its overly complex parameter settings and nonlinearity make it difficult to implement in practice.

[0005] Linear Active Disturbance Rejection Control (LADRC) greatly simplifies controller parameters through a bandwidth-based controller parameter tuning method. However, in traditional LADRC, the selection of the observer bandwidth becomes a performance bottleneck. If the bandwidth is too low, disturbance estimation lags; if the bandwidth is too high, sensor noise is amplified. A single LESO (Low-Frequency Active Disturbance Rejection) controller struggles to simultaneously handle low-frequency load disturbances and high-frequency switching noise.

[0006] To address the aforementioned issues, this invention designs a two-stage coupled extended state observer. This observer enables the identification of multi-frequency band disturbances in different zones, overcoming the contradiction between the operating state and bandwidth selection of the LADRC controller. It combines speed and robustness. When combined with the output characteristics of the QPR controller in the current inner loop, it reduces one coordinate transformation step, simplifies the control process, and achieves stronger robustness and stability of the grid-connected system. Summary of the Invention

[0007] To address the problems existing in the photovoltaic grid connection process, this invention proposes a dual closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer. By improving the voltage outer loop LADRC controller to enhance the stability and robustness of the DC bus voltage, and using a quasi-proportional resonant controller in the current inner loop to control the grid-connected current, this invention overcomes the shortcomings of existing technologies.

[0008] To achieve the above-mentioned objectives, the present invention adopts the following technical solution:

[0009] A dual-closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer includes the following steps:

[0010] Step 1: Establish the inverter mathematical model and determine the grid-connected current i along the d-axis. d It is the stable DC bus voltage U dc Control signals;

[0011] Step 2: Establish a novel LADRC controller based on a two-stage coupled extended state observer;

[0012] Step 3: Construct a high-performance dual-closed-loop grid-connected control system based on the LADRC controller and QPR controller with a two-level coupled extended state observer.

[0013] Furthermore, based on the inverter mathematical model, the key control variable, the grid-connected current i along the d-axis... d With the bus voltage U of the controlled object dc A two-stage coupled extended state observer is established, comprising an upper-stage LESO and a lower-stage LESO. The upper-stage LESO is responsible for estimating the state information of the control objective under steady-state conditions and filtering out low-frequency disturbances caused by high-frequency interference. make In conjunction with the lower-level LESO, dynamic information in the high-frequency domain is obtained, completing the integration of the total disturbance. The expression is:

[0014]

[0015] in, The tracking U output by the state observer in the low-frequency domain is respectively dc The value and its derivative k represents the estimated perturbation output by the upper-level LESO under low bandwidth conditions. 11 k 12 k 13 The feedback gain matrix is ​​formed, where x1 represents U. dc b is the compensation coefficient, and u represents the control signal, i.e., the obtained d-axis current reference value.

[0016] The lower-level LESO receives the state output from the upper-level LESO and uses the higher bandwidth to supplement the estimation of the system's remaining disturbances, expressed as:

[0017]

[0018] in, The tracking U output by the state observer in the high-frequency domain are respectively dc The value and its derivative To evaluate the noise estimation results in the high-frequency domain, k 21 k 22 k 23 Construct the feedback gain matrix;

[0019] Given the observed state, designing the LSEF linear control law yields the following virtual control signal:

[0020]

[0021] Where u0 is the virtual control signal, K p K d For the gain of the controller, DC bus voltage U dc The set value, This is the result of weighting the state estimates of the upper-level LESO and the lower-level LESO, i.e.:

[0022]

[0023] Wherein, ω1 and ω2 are weighting functions, which assign weights to different state signal components in a specific frequency domain;

[0024] u0 is a virtual control signal. The LADRC controller has the ability to automatically eliminate disturbances, and its control signal is:

[0025]

[0026] Furthermore, the LADRC controller used in the outer voltage loop of the photovoltaic grid-connected system derives the second-order differential equation of the DC bus voltage from the inverter mathematical model, and then reorganizes it into an integral series form:

[0027]

[0028] Among them, i d * The obtained grid-connected current reference value is the control signal of the LADRC controller;

[0029] The above equation can be written in the form of a mathematical model for an LADRC controller:

[0030]

[0031] Wherein, the state variable f is the total system disturbance;

[0032] Transforming the above equation into state-space form:

[0033]

[0034] Where x1 and x2 represent the DC bus voltage U dc and its derivative x3 represents the total disturbance f of the system; v is the derivative of f;

[0035] The upper-level LESO establishes a third-order LESO expression based on the voltage differential equation. This involves setting state variables to follow the system state and disturbances, and then estimating these variables by adding appropriate gain coefficients. The expression is as follows:

[0036]

[0037] The estimated total disturbance is fed into the next level LESO to obtain estimates of high-frequency disturbances and reference values:

[0038]

[0039] Thus, the total perturbation is obtained.

[0040] Design a linear error feedback law to obtain the control signal u of the LADRC controller, i.e.

[0041]

[0042] Set q-axis current reference value To maximize the grid-connected power factor, the actual dq-axis current of the system is strongly coupled and cannot be directly controlled. Therefore, the Parker inverse transformation is used to convert it into an AC signal in a two-phase stationary coordinate system, namely the αβ-axis current, and the AC signal is directly controlled.

[0043] Using a QPR controller in the inner current loop, a damping term is added to the denominator of the PR controller's transfer function, transforming the PR controller's infinite gain at the target frequency into a finite gain near the target frequency. The QPR controller can then achieve tracking control of the input signal, and its transfer function G... QPR (s) is:

[0044]

[0045] Among them, K p K is the proportional element coefficient, which affects the amplitude other than the controller's resonant frequency; R ω is the resonance factor, which affects the gain of the controller across the entire frequency range; c The cutoff frequency determines the control bandwidth, ω. c The larger the value, the larger the controller bandwidth; ω0 is the resonant frequency.

[0046] From the mathematical model of the grid-connected inverter in the two-phase stationary coordinate system, we can obtain:

[0047]

[0048] Among them, e α e β For the description of the grid-side voltage in a two-phase stationary coordinate system, i α i β Let u be the inductor current in a two-phase stationary coordinate system. α u β i is the inverter output voltage in a two-phase stationary coordinate system; α i β The reference signal obtained through the voltage outer loop The difference is sent to the QPR controller; the control signal is then transmitted. With the measured grid-side voltage e α e β The reference voltage vector is obtained by taking the difference, i.e.:

[0049]

[0050] in, This is the control signal for the current inner loop QPR controller. The voltage reference vector is used to control the switching devices via the SVPWM module to complete the inversion.

[0051] The beneficial technical effects of this invention are as follows:

[0052] This invention designs a two-stage coupled extended state observer, which can design extended state observers with different bandwidths for state signals and disturbance signals in different frequency ranges, thereby achieving effective estimation of signals at different frequencies. The estimated state information and total disturbance are weighted and processed as a feedback signal to be sent to the LADRC controller, greatly improving the robustness and stability of the system. To address the higher dynamic response requirements of the inner current loop, a QPR controller is designed, which reduces one coordinate transformation while ensuring improved grid-connected current power factor, enabling more agile current control. This method has strong adaptability and engineering feasibility, and is particularly suitable for scenarios with multiple inverters operating in parallel and high power quality requirements, possessing broad application and promotion value. Attached Figure Description

[0053] Figure 1 This is the main circuit topology of the two-level inverter on the grid side in this invention;

[0054] Figure 2 This is a structural diagram of the LADRC controller with a two-stage coupled extended state observer in a grid-connected environment according to the present invention;

[0055] Figure 3 This is a structural diagram of the dual closed-loop grid-connected control system in this invention;

[0056] Figure 4 The simulation results of this invention are shown in the figure. Detailed Implementation

[0057] The specific embodiments of the present invention will be further described below with reference to specific examples:

[0058] A dual-closed-loop control method for grid-connected inverters based on a two-stage coupled extended state observer includes the following steps:

[0059] Step 1: Establish a mathematical model of the inverter. Based on the mathematical model and the working principle of the inverter, determine the grid-connected current i along the d-axis. d It is the stable DC bus voltage U dc The control quantity;

[0060] Figure 1 The main circuit topology of the grid-side two-level inverter is shown. The full-bridge circuit uses controllable switching devices (IGBTs) with anti-parallel diodes as freewheeling diodes. A switching function is introduced throughout the control process:

[0061]

[0062] The mathematical model of the inverter can be obtained from Kirchhoff's voltage law and current law, as shown in equation (2):

[0063]

[0064] Among them, e k For grid-side load, i is the grid voltage. sk For grid-connected current, u kN Let k = a, b, c, representing the voltage from each phase to the common point; u NO I is the voltage from the load to the common point. dc U dc For the DC side output current and output voltage; I L This represents the current flowing through the inductor side.

[0065] The operating mode of the IGBT affects the voltage levels, namely:

[0066]

[0067] Then we have:

[0068]

[0069] The mathematical model of the grid-connected inverter under a three-phase AC environment is as follows:

[0070]

[0071] Where L is the filter inductance, R is the associated resistance, C is the capacitance, and i sa i sb i sc e represents the actual three-phase current flowing through the filter. a e b e c For network-side load, S a S b S c For the switching function, U dc For DC voltage source, i a i b i c This refers to the three-phase current of the inverter-side bridge arm;

[0072] To reduce the complexity of controller design, the motion parameters in the three-phase stationary coordinate system are transformed into relatively stationary parameters in the two-phase rotating coordinate system, i.e., the dq coordinate system, using the following expression:

[0073]

[0074] Among them, i d i q For the description of the grid-side current in a two-phase rotating coordinate system, u d u q This describes the voltage on the bridge arm side in a two-phase rotating coordinate system, where ω is the angular velocity and e is the voltage on the bridge arm side. d e q S is a description of the grid-side voltage in a two-phase rotating coordinate system. d S q Let I be the equivalent switching function in a two-phase rotating coordinate system, and let I0 be the equivalent power supply output current in a two-phase rotating coordinate system.

[0075] In the dq coordinate system, the component of the grid voltage mapped onto the d-axis is |E|, and the component onto the q-axis is 0; according to instantaneous power theory, the active power P and reactive power Q output by the inverter to the grid are:

[0076]

[0077] And because of e d =|E|,e q =0, so u d and i d This is related to the active power of the control system; and since the actual active power P of the photovoltaic grid-connected inverter system is provided by the DC power supply, under the condition of neglecting the power loss of power electronic devices, it can be considered that P = P dcTherefore, the DC bus voltage can be controlled by controlling the d-axis current; that is, with a constant power, adjusting i... d It can indirectly control U dc .

[0078] In summary, based on the inverter's mathematical model and its working principle analysis, this invention selects the grid-connected current i along the d-axis. d As a control variable, by adjusting i d To ensure bus voltage U dc Stability.

[0079] Step 2: Establish a novel LADRC controller based on a two-stage coupled extended state observer;

[0080] The main function of the extended state observer in this invention is to estimate various disturbances that may exist in the grid-connected inverter system and to counteract these disturbances by designing control laws. Since both high-frequency and low-frequency disturbances exist in real-world grid-connected inverter systems, a single-bandwidth LESO cannot effectively perform its task. This invention designs a two-stage coupled extended state observer, specifically different LESOs for high-frequency and low-frequency disturbances, and cascades these different LESOs. Specifically, the state information and low-frequency disturbances are first estimated using a low-frequency LESO. Then, the disturbances are treated as known state information and fed into the next-level LESO, allowing the disturbances and reference quantities to be used together as known conditions to obtain the state information and disturbances in the high-frequency domain. Finally, the state variables obtained in the above process are weighted according to the requirements of the actual environment. The weighting function distributes the weights of the state variables estimated by the upper and lower-level LESOs, achieving a reasonable division of state variables corresponding to different types of disturbances.

[0081] Based on this, the present invention further designs an LADRC controller with a two-stage coupled extended state observer, such as... Figure 2 As shown, it includes upper-level LESO and lower-level LESO.

[0082] Although the first derivative of the DC bus voltage with respect to time can reflect its rate of change, the system's response speed often fails to meet the requirements of real-time regulation in the face of rapid disturbances to the bus voltage caused by adverse external conditions. From (5) and (6), the second-order differential equation of the DC bus voltage is obtained, which allows for accurate evaluation of various types of disturbances affecting the DC bus voltage and compensation through the controller;

[0083] The integral series form of the second-order differential equation for the DC bus voltage is:

[0084]

[0085] Among them, i d *The obtained grid-connected current reference value is the control signal of the LADRC controller;

[0086] The above equation can be written in the form of a mathematical model for an LADRC controller:

[0087]

[0088] Wherein, the state variable f is the total system disturbance;

[0089] Transforming the above equation into state-space form:

[0090]

[0091] Where x1 and x2 represent the DC bus voltage U dc and its derivative x3 represents the total disturbance f of the system; v is the derivative of f;

[0092] Traditional extended state observers (LESOs) exhibit good estimation and suppression capabilities when faced with only a single type of disturbance, such as purely low-frequency grid disturbances or purely high-frequency switching noise. However, in actual grid-connected operation, systems often encounter multi-frequency disturbances simultaneously, such as low-frequency disturbances caused by changes in power frequency load, high-frequency noise, and even spike signals caused by non-ideal behavior of switching devices. Engineering experience shows that disturbances such as irradiance variations, load steps, and power pulsations exist in the 0-200 Hz frequency domain, while filter resonances and sensor noise can reach 500-1000 Hz. Therefore, a dual-band LESO can be used to handle disturbances in different frequency domains.

[0093] The state variable estimated by the third-order LESO is U. dc , And the total system disturbance, due to the upper-level LESO bandwidth ω o The values ​​are relatively low, thus effectively estimating low-frequency interference. When estimating high-frequency interference, since the lower-level LESO has a higher bandwidth, it can ensure effective tracking of high-frequency signals. Low-frequency interference can be directly estimated using the effective values ​​from the upper-level LESO, discarding the low-frequency disturbances in the power frequency range amplified in the lower-level LESO.

[0094] The function of the upper-level LESO is to estimate U in the low-frequency domain. dc and its derivative and disturbance make In conjunction with the lower-level LESO, dynamic information in the high-frequency domain is obtained, completing the integration of the total disturbance. The advantage of the upper-level LESO is that, by properly setting the pole positions, this level of observer provides robust state variable and low-frequency disturbance estimates for the system without causing oscillations in the observed signal. The expression is:

[0095]

[0096] in, The tracking U output by the state observer in the low-frequency domain is respectively dc The value and its derivative k represents the disturbance estimated by the upper-level LESO under low bandwidth conditions. 11 k 12 k 13 The gain is used to construct the feedback gain matrix, whose value is related to the observer bandwidth. The parameters are selected using the pole placement method in control theory. x1 represents U dc b is the compensation coefficient, and u represents the control signal, i.e., the obtained d-axis current reference value.

[0097] Based on equations (8) and (11), the matrix form of the third-order LESO is written, that is, the state variables are set to follow the system state and disturbance, and the estimation is completed by adding a reasonable gain coefficient to the state variables (the actual output of the system and the total disturbance). The expression is:

[0098]

[0099] The lower-level LESO receives state information estimated by the upper-level LESO and uses higher bandwidth to supplement the estimation of residual system disturbances (mainly high-frequency components). Since high bandwidth is sensitive to noise, to avoid repeated amplification of low-frequency interference, a disturbance suppression feedback mechanism estimated by the upper-level LESO is introduced in the lower-level LESO, thereby ensuring that high-frequency disturbance estimation is not contaminated by redundant low-frequency disturbances. The expression is:

[0100]

[0101] in, The tracking U output by the state observer in the high-frequency domain are respectively dc The value and its derivative For the observer to consider the estimated disturbance Under the premise of k, for the noise estimation results in the high-frequency domain 21 k 22 k 23 This forms the feedback gain matrix. Since the rate of change of the DC bus voltage in a grid-connected system is relatively small compared to high-frequency disturbances, the input information remains unchanged, and the state variables are estimated using the upper-level LESO. That is, in the lower-level LESO, the input information is still `bu`, but it is based on...

[0102] The lower-level LESO state-space expression is:

[0103]

[0104] Total disturbance is In this way, the parameters and disturbances in the system are clearly defined, and it can be approximated as an integral cascade type, which greatly simplifies the form of the control law.

[0105] The reason why the LADRC controller in this invention can use the LSEF linear control law is that the two-stage coupled state observer can effectively estimate the total disturbance of the system. The virtual control signal expression is designed as follows:

[0106]

[0107] Where u0 is the virtual control signal, K p and K d The gain of the controller is obtained by tuning using the pole placement method. DC bus voltage U dc The set value, This is the result of weighting the state variables estimated by the higher-level LESO and the lower-level LESO, i.e.:

[0108]

[0109] Wherein, ω1 and ω2 are weighting functions, which assign weights to different state signal components in a specific frequency domain;

[0110] The virtual control signal u0 is obtained according to formula (15). After disturbance compensation, the control signal u of the LADRC controller is obtained, that is...

[0111]

[0112] Ideally, due to the high sensitivity and foresight of high-bandwidth observers to system dynamic changes, extended state observers can capture and estimate disturbances in a timely manner before they are fully propagated to the system output and cause significant deviations. This ability to sense system fluctuations in advance means that even if high-frequency disturbances do not manifest as obvious anomalies in the output variables, they are identified by the observer as extended states (such as total disturbances) and transmitted to the controller. Therefore, for the system's state variables, the weights should be heavier on the state variables estimated by the upper-level LESO. The extended state variables are disturbances that utilize upper-level low-frequency disturbances and lower-level high-bandwidth conditional disturbances. Ultimately, the controller only receives a subset of key information after structural optimization, such as the combination of actual disturbances and target states, avoiding state variable redundancy. Based on this, the controller can apply compensating control inputs in advance, thereby effectively suppressing the impact of disturbances on system performance.

[0113] Step 3: Construct a high-performance dual-closed-loop grid-connected control system based on the LADRC controller and QPR controller with a two-level coupled extended state observer.

[0114] Although the voltage outer loop controller can directly generate control signals, it is limited by DC capacitor energy storage, A / D delay, and PWM delay. When the load on the grid-connected side is suddenly added or removed, the actual current is prone to deviate significantly from the reference. Relying solely on the outer loop compensation requires several fundamental cycles, resulting in excessive THD and transient peak values. Therefore, a high-speed current loop is needed to shorten the compensation time and increase the current limit. Thus, this invention constructs a high-performance dual-closed-loop grid-connected control system based on a two-stage coupled extended state observer LADRC controller and a QPR controller.

[0115] In this invention, the voltage outer loop of the dual closed-loop grid-connected control system adopts the novel LADRC controller described in step 2, the structure of which is as follows: Figure 3 As shown, the new LADRC outer loop controller obtains the input signal by comparing the DC bus voltage reference value with the actual value, and provides the grid-connected current reference signal. Settings Reference To maximize the grid-connected power factor, the actual dq-axis current of the system is strongly coupled, and the dq-axis current cannot be directly controlled. Therefore, the Parker inverse transformation is used to convert it into an AC signal in a two-phase stationary coordinate system, namely the αβ-axis current, and the AC signal is directly controlled.

[0116] Since the control signal provided by the voltage outer loop is an AC signal in a two-phase stationary coordinate system, the current inner loop uses a QPR controller, which facilitates zero-error tracking control of the AC signal.

[0117] In this invention, the inner current loop uses a QPR controller. A damping term is added to the denominator of the PR controller's transfer function, transforming the PR controller's infinite gain at the target frequency into a finite gain near the target frequency, significantly increasing the system's robustness and stability. Thanks to the PR controller's zero steady-state error tracking of AC signals, the QPR controller can achieve tracking control of the input signal, and its transfer function G... QPR (s) is:

[0118]

[0119] Among them, K p K is the proportional element coefficient, which affects the amplitude other than the controller's resonant frequency; R ξ is the resonance factor, affecting the controller's gain across the entire frequency range; ξ is the damping coefficient, suppressing the ∞dB peak to 40-60dB; enhancing robustness against ±1Hz grid frequency drift. c The cutoff frequency determines the controller's bandwidth, ω. c The larger the value, the larger the controller bandwidth; ω0 is the resonant frequency.

[0120] From the mathematical model of the grid-connected inverter in the two-phase stationary coordinate system, we can obtain:

[0121]

[0122] Among them, e α e β For the description of the grid-side voltage in a two-phase stationary coordinate system, i α i β Let u be the inductor current in a two-phase stationary coordinate system. α u β i is the inverter output voltage in a two-phase stationary coordinate system; α i β The reference signal obtained through the voltage outer loop The difference is fed to the QPR controller, that is:

[0123]

[0124] The reference signal passes through the proportional and resonant terms of the QPR controller. The proportional term K... p Provides full-band gain and increases the system's phase advance, improving phase margin; resonant term Additional gain is provided at a specific frequency (50Hz) to ensure zero steady-state error tracking at that frequency. Furthermore, due to the damping coefficient ξ, the resonant peak at this frequency can be adjusted by regulating the damping coefficient, balancing zero steady-state error and stability margin. The reference voltage output by the QPR controller and the actual sampled voltage are used to obtain the trigger signal for the SVPWM module.

[0125] control signal With the measured grid-side voltage e α e β The reference voltage vector is obtained by taking the difference, i.e.:

[0126]

[0127] in, This is the control signal for the current inner loop QPR controller. The voltage reference vector is used to control the switching devices via the SVPWM module to complete the inversion.

[0128] This invention employs a dual closed-loop architecture of "voltage outer loop + current inner loop." The outer loop is responsible for DC bus energy balance, while the inner loop uses a QPR controller for high-speed closed-loop tracking of the grid-connected current, achieving time-scale separation and model order reduction. The outer loop has low bandwidth and smooth regulation, ensuring the bus voltage remains stable under high-power fluctuations. The inner loop has high bandwidth, enabling rapid suppression of disturbances and limiting overcurrent. It utilizes the high gain of the QPR near the fundamental frequency to achieve zero steady-state error, while providing additional phase margin and virtual damping through proportional branches and damping parameters. This ensures the system maintains excellent dynamic robustness and low THD under parameter drift and harmonic interference conditions. The two loops work together, balancing rapid response, safety protection, and grid connection compliance, providing a highly efficient control solution that significantly improves inverter performance without increasing hardware costs.

[0129] After the system was built, the DC bus voltage reference value was set to 1250V. The system reliability was verified by simulating power fluctuations and load increases on the grid side at one second. Ideally, at the start of grid connection, the active power provided by the DC power supply equals the active power required by the load, and the grid power factor is maximized during the process, causing the DC bus voltage to quickly reach the target value. Due to the increase in load power requirement at 1 second, the power provided by the bus is insufficient, causing the bus voltage to drop. The photovoltaic modules then adjust the voltage level through power tracking, increasing the active power supply to restore the DC voltage. The experimental results are as follows: Figure 4 As shown, it meets the control objectives.

[0130] Compared to traditional LADRC control methods, this invention offers significant advantages in DC bus voltage control. By designing a two-stage coupled extended state observer structure, hierarchical estimation and feedback are performed for disturbances of different frequencies, effectively resolving the contradiction between "response speed and noise suppression" inherent in traditional LADRC bandwidth selection. The upper-stage LESO focuses on robust estimation of low-frequency disturbances, while the lower-stage LESO utilizes high bandwidth for rapid response to high-frequency disturbances. A disturbance feedforward suppression mechanism is introduced into the structure to avoid repeated disturbance estimation and amplification, significantly improving the system's robustness and dynamic response performance. Thanks to the rapid disturbance estimation, the QPR controller used in the inner current loop can quickly track the reference current. Furthermore, since the QPR controller can directly control AC quantities in the two-phase stationary coordinate system, control efficiency is greatly improved. The two loops complement each other, jointly completing the control of the grid-connected inverter.

[0131] Of course, the above description is not intended to limit the present invention, and the present invention is not limited to the examples given above. Any changes, modifications, additions or substitutions made by those skilled in the art within the scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A dual-closed-loop control method for a grid-connected inverter based on a two-stage coupled extended state observer, characterized in that, Includes the following steps: Step 1: Establish the mathematical model of the inverter and determine the grid-connected current under the d-axis. It is a stable DC bus voltage Control signals; Step 2: Establish an LADRC controller based on a two-stage coupled extended state observer; Step 3: Construct a high-performance dual-closed-loop grid-connected control system based on the LADRC controller and QPR controller with a two-stage coupled extended state observer; Based on the inverter mathematical model, the key control variable is the grid-connected current along the d-axis. DC bus voltage of the controlled object A two-stage coupled extended state observer is established, comprising an upper-stage LESO and a lower-stage LESO. The upper-stage LESO is responsible for estimating the state information of the control objective under steady-state conditions and filtering out low-frequency disturbances caused by high-frequency interference. ,make In conjunction with the lower-level LESO, dynamic information in the high-frequency domain is obtained, completing the integration of the total disturbance. The expression is: ； in, , The tracking output of the state observer in the low-frequency domain is respectively The value and its derivative , This represents the estimated perturbation output by the upper-level LESO under low bandwidth conditions. , , The feedback gain matrix is ​​formed. represent , For compensation coefficient, This represents the control signal, i.e., the obtained d-axis current reference value. ; The lower-level LESO receives the state output from the upper-level LESO and uses the higher bandwidth to supplement the estimation of the system's remaining disturbances, expressed as: ； in, , The tracking output of the state observer in the high-frequency domain is respectively The value and its derivative , For the noise estimation results in the high-frequency domain, , , Construct the feedback gain matrix; Given the observed state, designing the LSEF linear control law yields the following virtual control signal: ； in, For virtual control signals, , For the gain of the controller, DC bus voltage The set value, , This is the result of weighting the state estimates of the upper-level LESO and the lower-level LESO, i.e.: ； in, , The weighting function assigns weights to different state signal components in a specific frequency domain. As a virtual control signal, the LADRC controller has the ability to automatically eliminate disturbances, and its control signal is: 。 2. The dual closed-loop control method for a grid-connected inverter based on a two-stage coupled extended state observer as described in claim 1, characterized in that: The LADRC controller used in the outer voltage loop of the photovoltaic grid-connected system derives the second-order differential equation of the DC bus voltage from the inverter mathematical model, and then reorganizes it into an integral series form: ； in, The obtained grid-connected current reference value is the control signal of the LADRC controller; For filter inductance; For capacitors; For associated resistance; Angular velocity; The equivalent switching function in the d-axis of the two-phase rotating coordinate system; The equivalent switching function in the q-axis of the two-phase rotating coordinate system; This describes the grid-side voltage in the d-axis of a two-phase rotating coordinate system. This describes the grid-side current in the d-axis of a two-phase rotating coordinate system. This describes the voltage on the bridge arm side in the d-axis of a two-phase rotating coordinate system. The above equation can be written in the form of a mathematical model for an LADRC controller: ； Among them, state variables The total system disturbance; Transforming the above equation into state-space form: ； in, , Represents the DC bus voltage and its derivative , Represents the total disturbance of the system ; for The derivative; The upper-level LESO establishes a third-order LESO expression based on the voltage differential equation. This involves setting state variables to follow the system state and disturbances, and then estimating these variables by adding appropriate gain coefficients. The expression is as follows: ； The estimated total disturbance is fed into the next level LESO to obtain estimates of high-frequency disturbances and reference values: ； Thus, the total perturbation is obtained. ; Design a linear error feedback law to obtain the control signal for the LADRC controller. ,Right now : ； Set q-axis current reference value To maximize the grid-connected power factor, the actual dq-axis currents of the system are strongly coupled, making direct control of the dq-axis currents impossible. Therefore, the Parker inverse transform is used to convert the signals into AC signals in a two-phase stationary coordinate system, i.e. The shaft current directly controls this AC signal; Using a QPR controller in the inner current loop, a damping term is added to the denominator of the PR controller's transfer function, transforming the PR controller's infinite gain at the target frequency into a finite gain near the target frequency. The QPR controller can then achieve tracking control of the input signal, and its transfer function... for: ； in, This is the proportional element coefficient, which affects the amplitude other than the controller's resonant frequency; The resonant factor affects the gain of the controller across the entire frequency range; The cutoff frequency determines the control bandwidth. The larger the value, the greater the controller bandwidth; The resonant frequency; From the mathematical model of the grid-connected inverter in the two-phase stationary coordinate system, we can obtain: ； in, , This describes the grid-side voltage in a two-phase stationary coordinate system. , The inductor current is in a two-phase stationary coordinate system. , The inverter output voltage is in a two-phase stationary coordinate system. , The reference signal obtained through the voltage outer loop , The difference is sent to the QPR controller; the control signal is then transmitted. , With the measured grid-side voltage , The reference voltage vector is obtained by taking the difference, i.e.: ； in, , This is the control signal for the current inner loop QPR controller. , The voltage reference vector is used to control the switching devices via the SVPWM module to complete the inversion.

Images