A grid-connected inverter adaptive predictive control method and system

Abstract

A grid-connected inverter adaptive predictive control method and system, according to the discrete equivalent impedance model of the grid-connected inverter system cannot represent the unmodeled dynamics of the grid-connected inverter system, set the heterogeneous filter general compensation bias term, superimpose the heterogeneous filter general compensation bias term on the discrete equivalent impedance model, obtain the adaptive impedance model; the unmodeled dynamics of the grid-connected inverter system includes: model truncation error, grid-connected point voltage disturbance, LCL resonant loop impedance characteristic change, power device non-ideal characteristic; after updating the heterogeneous filter general compensation bias term according to the real-time data of the grid-connected inverter system, the parameters in the adaptive impedance model are identified online; based on the adaptive impedance model after parameter identification, taking the minimum error between the inverter bridge arm current and the corresponding expected value as the control target, the inverter output voltage reference value is determined to generate the driving pulse signal, and the dependence of the grid-connected inverter control algorithm on the filter topology and parameters is solved.

Description

Technical Field

[0001] This invention belongs to the field of power electronic converter technology, specifically, it relates to an adaptive predictive control method and system for grid-connected inverters that does not require prior parameters of filters. Background Technology

[0002] In recent years, the grid connection of renewable energy power generation such as wind power and photovoltaic power has developed rapidly. As the core interface between new energy power generation equipment and the power grid, the grid-connected inverter's control performance directly affects power quality, system stability, and grid standard compliance. To achieve low harmonic distortion grid-connected current, L-type, LC-type, or LCL-type power filters are usually connected between the inverter and the grid. In practical engineering applications, filter parameters can drift significantly due to factors such as temperature changes, magnetic saturation effects, device aging, and manufacturing tolerances, leading to "model mismatch" between the theoretical design model and the actual physical system. Furthermore, in pursuit of better filtering effects or due to cost and size constraints, different types of filters are often selected for different application scenarios, making it difficult for fixed-model controllers designed for a single topology to be universally applicable.

[0003] The aforementioned model mismatch and topology dependency issues pose a serious challenge to traditional high-performance control algorithms. First, model inaccuracies directly lead to decreased controller accuracy, resulting in increased steady-state tracking errors, aggravated current waveform distortion, and even threatening system stability when parameters deviate significantly. Second, adapting to different filters typically requires remodeling and tediously adjusting controller parameters, greatly increasing the cost and complexity of product development, maintenance, and upgrades. Although existing research has attempted to enhance system robustness through online parameter identification or state observation, these methods often still require prior knowledge of the filter topology and typically rely on additional sensors such as grid-side current or capacitor voltage, increasing system costs and introducing new points of failure. Summary of the Invention

[0004] To address the issue of the dependence of grid-connected inverter control algorithms on filter topology and parameters, this invention proposes an adaptive predictive control method and system for grid-connected inverters. The control method possesses inherent adaptive capabilities, enabling seamless adaptation to different types of filters without altering the control structure or performing complex parameter tuning. Simultaneously, it maintains robust and high-performance power control under various grid conditions using only a minimal number of sensors. This is not only crucial for improving the versatility and reliability of grid-connected inverter products but also a core technological guarantee for promoting the efficient and flexible grid connection of renewable energy.

[0005] The present invention adopts the following technical solution.

[0006] This invention proposes an adaptive predictive control method for a grid-connected inverter, wherein the AC side of the inverter is connected to the power grid via a filter, and the method includes: The inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system are obtained in a two-phase stationary coordinate system. The grid-connected inverter system includes an inverter and a filter, and the AC side of the inverter is connected to the grid through the filter. A discrete equivalent impedance model of the grid-connected inverter system is established using the difference between the inverter output voltage and the grid connection point voltage of the grid-connected inverter system, and the inverter arm current. Based on the unmodeled dynamics of the grid-connected inverter system that cannot be represented by the discrete equivalent impedance model, a general compensation bias term for heterogeneous filters is set and superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. The unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic change, and power device non-ideal characteristics. After updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system, the parameters in the adaptive impedance model are identified online. The inverter arm current reference value is determined based on the grid-side current reference value as the expected value of the inverter arm current; based on the adaptive impedance model after parameter identification, the inverter output voltage reference value is determined with the goal of minimizing the error between the inverter arm current and the corresponding expected value; and a drive pulse signal is generated according to the inverter output voltage reference value for adaptive predictive control of the inverter.

[0007] The steps for establishing a discrete equivalent impedance model of the grid-connected inverter system using the difference between the inverter output voltage and the grid connection point voltage of the grid-connected inverter system, and the inverter arm current, include: Within one control cycle, the inverter output voltage Grid connection point voltage of the grid-connected inverter system The difference is used as the input, with the inverter arm current as the input. For the output, based on the difference equation, a discrete equivalent impedance model of the grid-connected inverter system is established, as shown in the following equation:

[0008] In the formula, for Time in coordinate system Inverter bridge arm current, for Time in coordinate system -1 inverter bridge arm voltage, For a moment -1 Grid connection point voltage in coordinate system For the current term coefficient, This is the voltage term coefficient.

[0009] Based on the fact that the discrete equivalent impedance model cannot characterize the unmodeled dynamics of the grid-connected inverter system, a general compensation bias term for heterogeneous filters is set. This general compensation bias term is then superimposed on the discrete equivalent impedance model to obtain the adaptive impedance model. The steps include: Truncate the error bias term using the model Grid connection point voltage disturbance bias term Resonant impedance characteristic bias term and systematic bias term The weighted sum serves as a general compensation bias term for heterogeneous filters. The general compensation bias term of the heterogeneous filter is superimposed on the discrete equivalent impedance model to obtain the adaptive impedance model.

[0010] The calculation steps for the model truncation error bias term include: calculating the ideal value of the inverter arm current based on the discrete equivalent impedance model, obtaining the actual value of the inverter arm current by directly sampling through a current sensor, filtering the difference between the actual value and the ideal value of the inverter arm current to obtain the oscillating current component caused by the truncation error, and setting the model truncation error bias term based on the oscillating current component.

[0011] The calculation steps for the model truncation error bias term include: if the absolute value of the change in grid connection point voltage between the current control cycle and the previous control cycle is not less than the preset voltage threshold, then the change in grid connection point voltage is used as the grid connection point voltage disturbance bias term.

[0012] The calculation steps for the resonant impedance characteristic bias term include: filtering the difference between the actual and ideal values ​​of the inverter arm current to obtain the resonant current component at the resonant frequency; and setting the resonant impedance characteristic bias term based on the resonant current component.

[0013] The calculation steps for the systematic bias term include: constructing an observation equation for the systematic bias introduced by the non-ideal characteristics of the power device; identifying the parameters of the observation equation using the recursive least squares method; and constructing the systematic bias term using the identified parameters and the systematic bias.

[0014] The steps for superimposing the general compensation bias term of the heterogeneous filter onto the discrete equivalent impedance model to obtain the adaptive impedance model include: The adaptive impedance model of a grid-connected inverter system is shown in the following equation:

[0015] In the formula, This is a general compensation bias term for heterogeneous filters.

[0016] After updating the heterogeneous filter's general compensation bias term based on real-time data from the grid-connected inverter system, the steps for online identification of parameters in the adaptive impedance model include: With current term coefficient Voltage term coefficient Constructing the parameter vector to be identified ; Based on real-time data from the grid-connected inverter system, a recursive identification mechanism with dynamic updates of error covariance is used to identify the parameter vector online, thereby obtaining the parameter estimation vector for the current control cycle. , , They are time points Current term coefficient Voltage term coefficient The estimated value.

[0017] The steps for determining the inverter arm current reference value as the expected value of the inverter arm current based on the grid-side current reference value include: Based on the active power command and reactive power command from the power grid, and combined with the grid connection point voltage, the reference value of the inverter arm current is calculated as follows:

[0018]

[0019] In the formula, , for Time in coordinate system Inverter bridge arm current reference value, , They are time points The active power command and reactive power command of the power grid. , for Time in coordinate system Grid connection point voltage; Inverter bridge arm current reference value As a moment +2 inverter arm current expected value .

[0020] Based on the adaptive impedance model after parameter identification, and with the control objective of minimizing the error between the inverter arm current and its corresponding expected value, the steps for determining the inverter output voltage reference value include: The adaptive impedance model after parameter identification is used as a one-step forward predictor, with the control objective being to minimize the error in tracking the desired value of the inverter arm current. The time step is solved by algebraic back-calculation. The analytical expression for the +1 inverter output voltage reference value is:

[0021]

[0022] In the formula, , for Time in coordinate system +1 is the inverter output voltage reference value. , for Time in coordinate system +2 inverter arm current expected value , for Time in coordinate system +1 inverter bridge arm current, , They are time points Current term coefficient Voltage term coefficient The estimated value, For a moment The general compensation bias term for heterogeneous filters.

[0023] The steps for adaptive predictive control of the inverter, which involve generating drive pulse signals based on the inverter output voltage reference value, include: Time The inverter output voltage reference value of +1 is mapped to the three-phase stationary coordinate system through inverse coordinate transformation, and then normalized based on the DC bus voltage to obtain the continuous switching duty cycle vector, as shown in the following formula:

[0024] In the formula, For a moment The continuous switch duty cycle vector, Time in the three-phase stationary coordinate system The inverter output voltage reference value, This is the DC bus voltage.

[0025] The steps for adaptive predictive control of the inverter, which involve generating drive pulse signals based on inverter output voltage reference values, also include: Based on the finite switching state space of a two-level three-phase inverter, the continuous switching duty cycle vector is discretized and quantized using the nearest vector approximation principle to obtain the optimal switching state vector and generate the corresponding drive pulse signal.

[0026] In another aspect, the present invention also proposes an adaptive predictive control system for a grid-connected inverter, comprising: The model building module is used to obtain the inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system in a two-phase stationary coordinate system. The grid-connected inverter system includes an inverter and a filter, with the AC side of the inverter connected to the grid via the filter. A discrete equivalent impedance model of the grid-connected inverter system is established using the difference between the inverter output voltage and the grid connection point voltage, and the inverter arm current. Based on the unmodeled dynamics of the grid-connected inverter system that the discrete equivalent impedance model cannot represent, a general compensation bias term for heterogeneous filters is set and superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. The unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic changes, and non-ideal characteristics of power devices. The parameter identification module is used to identify the parameters in the adaptive impedance model online after updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system. The predictive control module is used to determine the inverter arm current reference value as the expected value of the inverter arm current based on the grid-side current reference value; based on the adaptive impedance model after parameter identification, the module determines the inverter output voltage reference value with the control objective of minimizing the error between the inverter arm current and the corresponding expected value; and generates a drive pulse signal according to the inverter output voltage reference value to perform adaptive predictive control of the inverter.

[0027] The present invention is also a terminal, including a processor and a storage medium; the storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the steps of the method.

[0028] The present invention is also a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method.

[0029] Compared with the prior art, the present invention has at least the following technical effects: the method proposed in this invention can realize adaptive high-precision predictive control of grid-connected inverters with arbitrary topology filters without relying on the prior parameters of the filter or changing the control structure.

[0030] This invention proposes a universal compensation bias term for heterogeneous filters, unifying the unmodeled dynamics with vastly different physical mechanisms into a single time-varying bias term, which is superimposed on the discrete equivalent impedance model. Without altering the structure of the first-order difference equation, this bias term absorbs all contributions from the unmodeled dynamics, enabling unified modeling of heterogeneous filters with arbitrary topologies without relying on filter topology information. Specifically, resonant components are extracted through current residual bandpass filtering, using their inverse as the bias to implicitly compensate for model truncation errors without requiring knowledge of the filter order. Threshold-triggered compensation based on voltage rate of change detection compensates for voltage feedforward delay effects in real time, with a response within one cycle. Real-time extraction and adaptive damping injection of resonant components implicitly suppress LCL resonance without adding an active damping controller. Online identification of three parameters—polarity correlation, current correlation, and global deviation—eliminates the need for offline measurement of dead zone and voltage drop parameters, and automatically adapts to temperature aging.

[0031] This invention also proposes a two-layer architecture for separating and identifying physical parameters and unmodeled dynamics, with the inner layer updating the bias term every cycle and the outer layer recursively updating the model parameters.

[0032] This invention achieves adaptive predictive control of grid-connected inverters that requires no prior information about the filter by using the same discrete equivalent impedance model (first-order difference equation), the same universal compensation bias for heterogeneous filters, and the same backward recursive control. It can seamlessly adapt to heterogeneous filters with any topology. Attached Figure Description

[0033] Figure 1 This is a flowchart of the adaptive predictive control method for grid-connected inverters proposed in this invention; Figure 2 This is a schematic diagram of the connection system between the grid-connected inverter and the power grid in an embodiment of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0035] This invention proposes an adaptive predictive control method for grid-connected inverters. The adaptive predictive control method does not require prior parameters of the filter. It achieves adaptive optimal control of the output current of the grid-connected inverter by constructing a discrete equivalent impedance model decoupled from the specific topology and parameters of the filter, and identifying the dynamic parameters of the decoupled model online.

[0036] like Figure 1 As shown, the method includes the following steps: Step 1: Establish a grid-connected inverter system, including an inverter and a filter. The AC side of the inverter is connected to the grid via the filter. Obtain the inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system in a two-phase stationary coordinate system.

[0037] In the embodiment, the connection system structure between the grid-connected inverter and the power grid is as follows: Figure 2 As shown, the established grid-connected inverter system includes a three-phase voltage source inverter and a coupling filter. The DC side of the three-phase voltage source inverter is connected to a DC voltage source. The AC side is connected to the power grid via a coupling filter. The coupling filter can be L-type, LC-type, LCL-type, or any other arbitrary topology, and its specific circuit form and electrical parameters (inductance, capacitance, resistance, etc.) are considered unknown during the controller design phase. This is the voltage at the grid connection point.

[0038] Time period Real-time acquisition of inverter arm voltage and current and grid connection point voltage; the inverter output voltage is defined in a three-phase stationary coordinate system (abc coordinate system) as... , , The inverter bridge arm current is , , The grid connection point voltage is , , Mapping electrical quantities from a three-phase stationary coordinate system to a two-phase stationary coordinate system ( (Coordinate system), using the constant amplitude Clarke transform, after transformation, we obtain Inverter output voltage in coordinate system Inverter bridge arm current and grid connection point voltage .

[0039] The advantage of using a two-phase stationary coordinate system instead of a rotating coordinate system is that it eliminates the need for a phase-locked loop (PLL) for grid phase tracking, avoiding the phase error and stability problems introduced by the PLL due to changes in grid impedance under weak grid conditions. At the same time, the control system structure under the two-phase stationary coordinate system is simpler, reducing the computational overhead of coordinate transformation and improving the real-time response speed of the controller.

[0040] Step 2: Using the difference between the inverter output voltage and the grid connection point voltage of the grid-connected inverter system, and the inverter arm current, establish a discrete equivalent impedance model of the grid-connected inverter system.

[0041] Specifically, within one control cycle, the inverter output voltage... Grid connection point voltage of the grid-connected inverter system The difference is used as the input, with the inverter arm current as the input. For the output, based on the difference equation, a discrete equivalent impedance model of the grid-connected inverter system is established, as shown in the following equation:

[0042] In the formula, for Time in coordinate system Inverter bridge arm current, for Time in coordinate system -1 inverter bridge arm voltage, For a moment -1 Grid connection point voltage in coordinate system For the current term coefficient, This is the voltage term coefficient.

[0043] In the embodiment, each moment corresponds to a control cycle, which meets the requirements of online predictive control.

[0044] The discrete equivalent impedance model of a grid-connected inverter system can provide a simple and accurate internal model for predictive control within the time-discrete framework of digital control. It is suitable for advanced algorithms such as predictive current control, deadbeat control, and active damping, and therefore serves as the basic model for adaptive predictive control of grid-connected inverters.

[0045] Moreover, the core point of the discrete equivalent impedance model of the grid-connected inverter system is that regardless of the topology of the coupling filter (L, LC, LCL, etc.), the transfer characteristics of the coupling filter can be expressed as a rational fractional impedance function when viewed from the inverter arm side port towards the grid side. Taking an LCL filter as an example, the transfer function As shown in the following formula:

[0046] In the formula, This refers to the inverter bridge arm current. This is the inverter output voltage. The voltage at the grid connection point. , This refers to the inverter-side inductance and its equivalent series resistance. , For the grid-side inductance and its equivalent series resistance, This is a filter capacitor.

[0047] However, when the filter topology is unknown, the order and coefficients of the continuous domain transfer function are uncertain and cannot be directly used for controller design. The discrete equivalent impedance model proposed in this invention fundamentally avoids this problem.

[0048] Step 3: Based on the unmodeled dynamics of the grid-connected inverter system that cannot be represented by the discrete equivalent impedance model, a general compensation bias term for heterogeneous filters is set, and the general compensation bias term for heterogeneous filters is superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. Among them, the unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic change, and power device non-ideal characteristics.

[0049] The discrete equivalent impedance model is a low-order difference equation that can only characterize the fundamental frequency and low-frequency dynamic characteristics of the filter. There are still many dynamic characteristics that cannot be characterized. Therefore, the discrete equivalent impedance model needs to be modified to accurately simulate the dynamic characteristics of coupled filters of different topologies.

[0050] Specifically, step 3 includes: Step 3.1: Calculate the ideal value of the inverter arm current based on the discrete equivalent impedance model, obtain the actual value of the inverter arm current by directly sampling through the current sensor, filter the difference between the actual value and the ideal value of the inverter arm current to obtain the oscillating current component caused by the truncation error, and set the model truncation error bias term based on the oscillating current component.

[0051] Since the filter topology is unknown, the dynamic characteristics of high-order filters cannot be fully characterized by first-order difference equations. The aforementioned first-order difference equations inevitably contain model truncation errors. The model truncation error caused by the truncation of the filter's high-order dynamics is essentially a structural model mismatch problem. That is, when a first-order L model is used to approximate a third-order LCL system, the dynamics of the resonant circuit are not described at all. The high-frequency residual modes of the high-order system generate low-frequency projections in the control band, which will seriously affect the closed-loop stability and dynamic performance. The truncation error is the quantitative expression of the high-frequency modes that are missing in the simplified model.

[0052] In the embodiment, in each control cycle, the ideal value of the inverter arm current is calculated based on the discrete equivalent impedance model. The actual value of the inverter arm current is obtained by directly sampling through a current sensor. Define current residual for and The difference, when there are unmodeled higher-order dynamics (such as LCL resonance), It will exhibit an oscillating waveform, with its oscillation frequency approximately equal to the resonant frequency of the LCL filter. As a direct manifestation of the model truncation error in the current domain, The signal is fed into a preset bandpass filter, and the output is the oscillating current component caused by the truncation error. ;based on The opposite number sets the model truncation error bias term. ; Truncate the error bias term in the model Superimposed on the discrete equivalent impedance model, the truncated higher-order dynamic contribution bias term is used to characterize the biased term. Absorbed.

[0053] Step 3.2: If the absolute value of the change in grid connection point voltage between the current control cycle and the previous control cycle is not less than the preset voltage threshold, then the change in grid connection point voltage is used as the grid connection point voltage disturbance bias term.

[0054] In the discrete equivalent impedance model, the grid connection point voltage only appears as the sampled value of the previous moment. However, in actual physical systems, the grid connection point voltage changes continuously and in real time. Moreover, as an external disturbance source, the grid connection point voltage directly affects the dynamic response of the inverter-side current. The disturbance effect of the grid voltage is not explicitly modeled in the discrete equivalent impedance model and needs to be compensated for by setting appropriate bias terms.

[0055] In this embodiment, the change in grid connection point voltage between the current control cycle and the previous control cycle is calculated. When the absolute value of the change in grid connection point voltage is less than a preset voltage threshold, the grid connection point voltage disturbance bias term is... The value is 0. When the absolute value of the change in grid connection point voltage is not less than the preset voltage threshold, the change in grid connection point voltage is used as the grid connection point voltage disturbance bias term. ; The grid connection point voltage disturbance bias term Superimposed on the discrete equivalent impedance model, the perturbation effect of the grid connection point voltage change on the current response is represented by the bias term. Absorbed.

[0056] Step 3.3: Filter the difference between the actual and ideal values ​​of the inverter bridge arm current to obtain the resonant current component at the resonant frequency; set the resonant impedance characteristic bias term based on the resonant current component.

[0057] When the filter is of type LCL, a resonant circuit exists consisting of inductors and capacitors. The impedance characteristics change drastically at the resonant frequency, and traditional controllers based on fixed parameters are prone to instability near this frequency. The bias term does not attempt to explicitly model the complete third-order dynamics of the LCL, but instead treats the influence of the resonant circuit as an equivalent time-varying, current-state-dependent disturbance bias current, which is superimposed on the original low-order model. This makes the output of the modified model as close as possible to the current response of the actual system, and can absorb the resonant component from the control channel regardless of how the resonant circuit changes.

[0058] In the embodiment, The signal is fed into a preset bandpass filter, and the resonant current component near the resonant frequency is output. The magnitude of this component reflects the strength of the current resonant energy, while its phase relationship reflects the phase difference between the resonant current and the excitation voltage. The opposite number sets the resonant impedance characteristic bias term. ; bias term of resonant impedance characteristic Superimposed on the discrete equivalent impedance model, the change in impedance characteristics of the LCL filter resonant circuit is represented by the bias term. Absorbed.

[0059] Step 3.4: Construct observation equations for the systematic bias introduced by the non-ideal characteristics of power devices, identify parameters of the observation equations using the recursive least squares method, and construct systematic bias offset terms using the identified parameters and systematic bias.

[0060] In practical engineering applications, the non-ideal characteristics of power devices (such as dead time effect, on-state voltage drop, switching delay, etc.) can also introduce systematic deviations in the bridge arm output voltage. Based on the discrete equivalent impedance model, these deviations are ultimately converted into current errors. Therefore, setting a bias term to unify all non-ideal factors into an equivalent bias current is automatically generated through online observation and learning. It is not necessary to know the specific values ​​of each non-ideal parameter explicitly. The bias term is used to compensate for these non-ideal factors, thereby improving the overall accuracy of the control system.

[0061] In this embodiment, the systematic biases introduced by the non-ideal characteristics of the power device include: compensating for fixed biases related to current polarity (such as dead zone effect, constant part of diode voltage drop), compensating for biases proportional to current magnitude (such as voltage drop caused by on-state resistance), and compensating for global biases unrelated to current (such as temperature drift, measurement zero drift, etc.). The observation equations are constructed using these three types of biases, and the parameters of the observation equations are identified using the recursive least squares method. The systematic bias term is then constructed using the identified parameters and the corresponding biases. ; The systematic bias term Superimposed on the discrete equivalent impedance model, the bias term characterizes the systematic deviation introduced by the non-ideal characteristics of the power device. Absorbed.

[0062] Step 3.5, truncate the error bias term using the model. Grid connection point voltage disturbance bias term Resonant impedance characteristic bias term and systematic bias term The weighted sum serves as a general compensation bias term for heterogeneous filters. The general compensation bias term of the heterogeneous filter is superimposed on the discrete equivalent impedance model to obtain the adaptive impedance model.

[0063] Heterogeneous filter general compensation bias term It is not a simple constant bias, but a time-varying adaptive compensation quantity that incorporates the combined effects of all unmodeled dynamics and external disturbances.

[0064] The adaptive impedance model of a grid-connected inverter system is shown in the following equation:

[0065] In this embodiment, the weight coefficient of each bias term is preferably 1, and the weight coefficient can be adjusted by an autonomous learning method based on the dynamic working condition perception results.

[0066] This invention, by setting a universal compensation bias term for heterogeneous filters, unifies the unmodeled dynamics with vastly different physical mechanisms, such as model truncation error, grid-connected point voltage disturbance, LCL resonant circuit impedance characteristic changes, and power device non-ideal characteristics, into several online-updable or real-time-calculated bias components, superimposed on the model output, without altering the core structure (first-order difference equation) of the discrete equivalent impedance model. This bias term allows the same discrete equivalent impedance model to seamlessly adapt to different output filter topologies, such as L-type, LCL, LLCL, and active filter types, without modifying the control algorithm structure, retuning controller parameters for different filters, or adding additional active damping loops or sensors. The adaptive update mechanism can track the equivalent contribution of resonant dynamics in real time, implicitly suppressing resonant peaks without introducing an additional active damping controller.

[0067] Step 4: After updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system, the parameters in the adaptive impedance model are identified online.

[0068] Specifically, using the current term coefficient Voltage term coefficient Constructing the parameter vector to be identified .

[0069] In the adaptive impedance model established in this invention, the physical parameters of the model itself are kept consistent with the actual system through online parameter identification, while the general compensation bias term of the heterogeneous filter always maintains the function of compensating for the unmodeled dynamics. Thus, a two-layer adaptive architecture of parameter identification and unmodeled compensation is formed, which can seamlessly adapt to different types of filters without changing the control structure or performing complex parameter tuning.

[0070] Specifically, based on real-time data from the grid-connected inverter system, a recursive identification mechanism with dynamic updates of error covariance is used to identify the parameter vector online and obtain the parameter estimation vector for the current control cycle. The specific execution process is as follows: Build Time regression vector ; Calculate the adaptive gain matrix ,in, For a moment The error covariance matrix is ​​-1. As a weighting factor, It is the identity matrix; Calculate prior prediction error ,in, For a moment -1 parameter estimation vector; Update the parameter estimation vector using the adaptive gain matrix and prior prediction error. ; Update error covariance matrix ;in, These are real-time estimates of the model parameters.

[0071] By executing step 4 in each control cycle, the adaptive impedance model can be updated in real time.

[0072] Step 5: Determine the inverter arm current reference value based on the grid-side current reference value as the inverter arm current expectation value; based on the adaptive impedance model after parameter identification, determine the inverter output voltage reference value with the control objective of minimizing the error between the inverter arm current and the corresponding expectation value; generate a drive pulse signal according to the inverter output voltage reference value for adaptive predictive control of the inverter.

[0073] Specifically, step 5 includes: Step 5.1: Based on the active power command and reactive power command from the power grid, and combined with the grid connection point voltage, calculate the reference value of the inverter arm current, as shown in the following formula:

[0074]

[0075] In the formula, , for Time in coordinate system Inverter bridge arm current reference value, , They are time points The active power command and reactive power command of the power grid. , for Time in coordinate system The grid connection point voltage.

[0076] In a typical digitally controlled grid-connected inverter, there is an inherent two-step delay from sampling to voltage output. To align the control target with the actual action time, the calculated inverter arm current reference value is used... Directly as time +2 inverter arm current expected value ,Right now = .

[0077] Step 5.2: Using the parameter-identified adaptive impedance model as a one-step forward predictor, with the control objective of minimizing the error between the inverter arm current and its corresponding expected value, the timing is solved by algebraic back-calculation. The analytical expression for the +1 inverter output voltage reference value is:

[0078]

[0079] In the formula, , for Time in coordinate system +1 is the inverter output voltage reference value. , for Time in coordinate system +2 inverter arm current expected value , for Time in coordinate system +1 inverter bridge arm current, , They are time points Current term coefficient Voltage term coefficient The estimated value, For a moment The general compensation bias term for heterogeneous filters.

[0080] Step 5.3, set the time... The inverter output voltage reference value of +1 is mapped to the three-phase stationary coordinate system through inverse coordinate transformation, and then normalized based on the DC bus voltage to obtain the continuous switching duty cycle vector.

[0081]

[0082] In the formula, For a moment The continuous switch duty cycle vector, Time in the three-phase stationary coordinate system The inverter output voltage reference value, This is the DC bus voltage.

[0083] Step 5.4: Based on the finite switching state space of the two-level three-phase inverter, the continuous switching duty cycle vector is discretized and quantized using the nearest vector approximation principle to directly lock the optimal switching state vector and generate the corresponding drive pulse signal.

[0084] Specifically, the components of the continuous switching duty cycle vector are discretized based on the limited set of 8 discrete switching states of a two-level three-phase inverter using a rounding quantization rule. Finally, the physically realizable switching state vector that is closest to the theoretically calculated value is selected as the optimal switching state vector. This process can actually be achieved by calculating the switching action time using techniques such as SPWM or SVPWM.

[0085] Based on the identified adaptive impedance model, this invention designs a back-calculation optimization control algorithm with the goal of minimizing the error of the inverter arm current tracking expectation. It solves for the desired inverter output voltage vector, and then generates the optimal switching state vector, which is applied to the power switches of each phase arm of the inverter to generate the corresponding pulse width modulation signal and control the voltage output of the inverter.

[0086] Each step of the adaptive predictive control method for grid-connected inverters proposed in this invention is written into executable code and run on a processor to form an electronic device that controls the grid-connected inverter. When the grid-connected inverter output uses an L-type, LC-type, or LCL-type filter, the method disclosed in this invention can achieve stable grid-connected power and current control.

[0087] In another aspect, the present invention also proposes an adaptive predictive control system for a grid-connected inverter, comprising: The model building module is used to obtain the inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system in a two-phase stationary coordinate system. The grid-connected inverter system includes an inverter and a filter, with the AC side of the inverter connected to the grid via the filter. A discrete equivalent impedance model of the grid-connected inverter system is established using the difference between the inverter output voltage and the grid connection point voltage, and the inverter arm current. Based on the unmodeled dynamics of the grid-connected inverter system that the discrete equivalent impedance model cannot represent, a general compensation bias term for heterogeneous filters is set and superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. The unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic changes, and non-ideal characteristics of power devices. The parameter identification module is used to identify the parameters in the adaptive impedance model online after updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system. The predictive control module is used to determine the inverter arm current reference value as the expected value of the inverter arm current based on the grid-side current reference value; based on the adaptive impedance model after parameter identification, the module determines the inverter output voltage reference value with the control objective of minimizing the error between the inverter arm current and the corresponding expected value; and generates a drive pulse signal according to the inverter output voltage reference value to perform adaptive predictive control of the inverter.

[0088] This disclosure can be a system, method, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for causing a processor to implement various aspects of this disclosure.

[0089] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0090] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0091] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0092] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. An adaptive predictive control method for a grid-connected inverter, characterized in that, include: The inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system are obtained in a two-phase stationary coordinate system. The grid-connected inverter system includes an inverter and a filter, and the AC side of the inverter is connected to the grid through the filter. A discrete equivalent impedance model of the grid-connected inverter system is established using the difference between the inverter output voltage and the grid connection point voltage of the grid-connected inverter system, and the inverter arm current. Based on the unmodeled dynamics of the grid-connected inverter system that cannot be represented by the discrete equivalent impedance model, a general compensation bias term for heterogeneous filters is set and superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. The unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic change, and power device non-ideal characteristics. Among them, the model truncation error bias term Grid connection point voltage disturbance bias term Resonant impedance characteristic bias term and systematic bias term The weighted sum serves as a general compensation bias term for heterogeneous filters. ; The calculation steps for the model truncation error bias term include: calculating the ideal value of the inverter arm current based on the discrete equivalent impedance model; obtaining the actual value of the inverter arm current by directly sampling through a current sensor; filtering the difference between the actual value and the ideal value of the inverter arm current to obtain the oscillating current component caused by the truncation error; and setting the model truncation error bias term based on the oscillating current component. The calculation steps for the grid connection point voltage disturbance bias term include: if the absolute value of the change in grid connection point voltage between the current control cycle and the previous control cycle is not less than the preset voltage threshold, then the change in grid connection point voltage is used as the grid connection point voltage disturbance bias term. The calculation steps for the resonant impedance characteristic bias term include: filtering the difference between the actual and ideal values ​​of the inverter arm current to obtain the resonant current component at the resonant frequency; and setting the resonant impedance characteristic bias term based on the resonant current component. The calculation steps for the systematic bias term include: constructing an observation equation for the systematic bias introduced by the non-ideal characteristics of the power device, identifying the parameters of the observation equation using the recursive least squares method, and constructing the systematic bias term using the identified parameters and the systematic bias. After updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system, the parameters in the adaptive impedance model are identified online. The inverter arm current reference value is determined based on the grid-side current reference value as the expected value of the inverter arm current; based on the adaptive impedance model after parameter identification, the inverter output voltage reference value is determined with the goal of minimizing the error between the inverter arm current and the corresponding expected value; and a drive pulse signal is generated according to the inverter output voltage reference value for adaptive predictive control of the inverter.

2. The adaptive predictive control method for grid-connected inverters according to claim 1, characterized in that, The steps for establishing a discrete equivalent impedance model of the grid-connected inverter system using the difference between the inverter output voltage and the grid connection point voltage of the grid-connected inverter system, and the inverter arm current, include: Within one control cycle, the inverter output voltage Grid connection point voltage of the grid-connected inverter system The difference is used as the input, with the inverter arm current as the input. For the output, a discrete equivalent impedance model of the grid-connected inverter system is established based on the difference equation.

3. The adaptive predictive control method for grid-connected inverters according to claim 2, characterized in that, The discrete equivalent impedance model of the grid-connected inverter system is shown in the following equation: In the formula, for Time in coordinate system Inverter bridge arm current, for Time in coordinate system -1 inverter bridge arm voltage, For a moment -1 Grid connection point voltage in coordinate system For the current term coefficient, This is the voltage term coefficient.

4. The adaptive predictive control method for grid-connected inverters according to claim 3, characterized in that, The steps for superimposing the general compensation bias term of the heterogeneous filter onto the discrete equivalent impedance model to obtain the adaptive impedance model include: The adaptive impedance model of a grid-connected inverter system is shown in the following equation: In the formula, This is a general compensation bias term for heterogeneous filters.

5. The adaptive predictive control method for grid-connected inverters according to claim 4, characterized in that, The step of online identification of parameters in the adaptive impedance model after updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system includes: With current term coefficient Voltage term coefficient Constructing the parameter vector to be identified ; Based on real-time data from the grid-connected inverter system, a recursive identification mechanism with dynamic updates of error covariance is used to identify the parameter vector online, thereby obtaining the parameter estimation vector for the current control cycle. , , They are time points Current term coefficient Voltage term coefficient The estimated value.

6. The adaptive predictive control method for grid-connected inverters according to claim 1, characterized in that, The step of determining the inverter arm current reference value as the expected value of the inverter arm current based on the grid-side current reference value includes: Based on the active power command and reactive power command from the power grid, and combined with the grid connection point voltage, the reference value of the inverter arm current is calculated as follows: In the formula, , for Time in coordinate system Inverter bridge arm current reference value, , They are time points The active power command and reactive power command of the power grid. , for Time in coordinate system Grid connection point voltage, inverter arm current reference value As a moment +2 inverter arm current expected value .

7. The adaptive predictive control method for grid-connected inverters according to claim 6, characterized in that, The adaptive impedance model based on parameter identification, with the control objective of minimizing the error between the inverter arm current and its corresponding expected value, includes the following steps in determining the inverter output voltage reference value: The adaptive impedance model after parameter identification is used as a one-step forward predictor, with the control objective being to minimize the error between the inverter arm current and its corresponding expected value. The time step is solved by algebraic back-calculation. The analytical expression for the +1 inverter output voltage reference value is: In the formula, , for Time in coordinate system +1 is the inverter output voltage reference value. , for Time in coordinate system +2 inverter arm current expected value , for Time in coordinate system +1 inverter bridge arm current, , They are time points Current term coefficient Voltage term coefficient The estimated value, For a moment The general compensation bias term for heterogeneous filters.

8. The adaptive predictive control method for grid-connected inverters according to claim 7, characterized in that, The step of generating a drive pulse signal based on the inverter output voltage reference value and performing adaptive predictive control of the inverter includes: Time The inverter output voltage reference value of +1 is mapped to the three-phase stationary coordinate system through inverse coordinate transformation, and then normalized based on the DC bus voltage to obtain the continuous switching duty cycle vector, as shown in the following formula: In the formula, For a moment The continuous switch duty cycle vector, Time in the three-phase stationary coordinate system The inverter output voltage reference value, This is the DC bus voltage.

9. The adaptive predictive control method for grid-connected inverters according to claim 8, characterized in that, The step of generating a drive pulse signal based on the inverter output voltage reference value and performing adaptive predictive control of the inverter further includes: Based on the finite switching state space of a two-level three-phase inverter, the continuous switching duty cycle vector is discretized and quantized using the nearest vector approximation principle to obtain the optimal switching state vector and generate the corresponding drive pulse signal.

10. An adaptive predictive control system for a grid-connected inverter, used to implement the adaptive predictive control method for a grid-connected inverter as described in any one of claims 1 to 9, characterized in that, include: The model building module is used to obtain the inverter output voltage, inverter arm current, and grid connection point voltage of the grid-connected inverter system in a two-phase stationary coordinate system. The grid-connected inverter system includes an inverter and a filter, with the AC side of the inverter connected to the grid via the filter. A discrete equivalent impedance model of the grid-connected inverter system is established using the difference between the inverter output voltage and the grid connection point voltage, and the inverter arm current. Based on the unmodeled dynamics of the grid-connected inverter system that the discrete equivalent impedance model cannot represent, a general compensation bias term for heterogeneous filters is set. This general compensation bias term is superimposed on the discrete equivalent impedance model to obtain an adaptive impedance model. The unmodeled dynamics of the grid-connected inverter system include: model truncation error, grid connection point voltage disturbance, LCL resonant circuit impedance characteristic changes, and non-ideal characteristics of power devices. The model truncation error bias term is used as the basis for this model. Grid connection point voltage disturbance bias term Resonant impedance characteristic bias term and systematic bias term The weighted sum serves as a general compensation bias term for heterogeneous filters. ; The ideal value of the inverter arm current is calculated based on the discrete equivalent impedance model; the actual value of the inverter arm current is obtained by direct sampling through a current sensor; the difference between the actual value and the ideal value of the inverter arm current is filtered to obtain the oscillating current component caused by the truncation error; and the truncation error bias term of the model is set based on the oscillating current component. If the absolute value of the change in grid connection point voltage between the current control cycle and the previous control cycle is not less than the preset voltage threshold, then the change in grid connection point voltage is used as the grid connection point voltage disturbance bias term. The difference between the actual and ideal values ​​of the inverter arm current is filtered to obtain the resonant current component at the resonant frequency; the resonant impedance characteristic bias term is set based on the resonant current component. The observation equation is constructed to account for the systematic bias introduced by the non-ideal characteristics of power devices. The parameters of the observation equation are identified by the recursive least squares method. The systematic bias term is constructed using the identified parameters and the systematic bias. The parameter identification module is used to identify the parameters in the adaptive impedance model online after updating the general compensation bias term of the heterogeneous filter based on the real-time data of the grid-connected inverter system. The predictive control module is used to determine the inverter arm current reference value as the expected value of the inverter arm current based on the grid-side current reference value; based on the adaptive impedance model after parameter identification, the module determines the inverter output voltage reference value with the control objective of minimizing the error between the inverter arm current and the corresponding expected value; and generates a drive pulse signal according to the inverter output voltage reference value to perform adaptive predictive control of the inverter.

11. A terminal, comprising a processor and a storage medium; characterized in that: The storage medium is used to store instructions; The processor is configured to operate according to the instructions to perform the steps of the method according to any one of claims 1-9.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the program implements the steps of the method according to any one of claims 1-9.

Images