A Harmonic Impedance Identification Method Based on Transient Energy Triggering and Covariance Correction

Abstract

This invention proposes a harmonic impedance identification method based on transient energy triggering and covariance correction, belonging to the technical field of harmonic impedance identification in AC distribution networks. The method includes: establishing a linear regression model of harmonic impedance and constructing a limit convergence model based on broadband voltage and current signals from the transmission line's point of common coupling; constructing a local transient energy variation index based on the broadband voltage signal, establishing a comparison relationship between this index and an adaptive threshold, and generating a binary state trigger signal; responding to this signal, establishing a frequency-band differentiated dynamic covariance adaptive correction Q matrix, superimposing this Q matrix onto the prior error covariance time update equation, modifying the recursive relationship of the covariance matrix, and embedding a piecewise robust weight function into the Kalman gain equation to perform harmonic impedance parameter updates; performing matrix trace and gain limit operations on the updated parameters, and outputting the broadband harmonic impedance identification result. This invention achieves rapid tracking of impedance abrupt changes and online identification of broadband harmonic impedance.

Description

Technical Field

[0001] This invention belongs to the field of harmonic impedance identification technology for AC distribution networks, and specifically relates to a harmonic impedance identification method based on transient energy triggering and covariance correction. Background Technology

[0002] With the increasing development of new power systems and the rise of high-proportion power electronic devices, inverter-type resources are gradually replacing synchronous machines as the main power source. Their phase-locked loop (PLL) and voltage-current dual-loop control structures introduce strong frequency domain coupling characteristics, leading to strong time-varying, nonlinear, and frequency-band-differentiated wideband harmonic impedance. Accurate online identification of harmonic impedance is a core fundamental technology for suppressing wideband oscillations, avoiding harmonic resonances, and optimizing damping control. Recursive least squares (RLS) is an important online parameter identification algorithm in industry due to its simple recursive structure and low computational cost. To meet the high steady-state identification accuracy requirements of power systems, high forgetting factor configurations are commonly used in engineering. However, existing technologies have fundamental defects that cannot be overcome.

[0003] Under the high forgetting factor setting in the recursive least squares method, the RLS error covariance matrix monotonically converges to a zero matrix as the number of sampling steps increases, the Kalman gain synchronously approaches zero, and the algorithm parameter update stagnates. When inverter resources undergo changes such as switching between grid-type and grid-following control modes, line switching, or load surges causing impedance step changes, the algorithm cannot accurately respond to parameter changes, leading to distorted identification results. This is an inherent mathematical defect of the RLS algorithm. Furthermore, because the system's broadband harmonic impedance is affected by frequency bands, harmonic impedances at different frequency bands exhibit different impedance characteristics in response to changes in system conditions. This further increases the requirements for accurate and timely identification of system-side harmonic impedance.

[0004] Therefore, the existing improved recursive least squares (RLS) method has shortcomings in mathematical rigor, frequency band decoupling capability, and accuracy loss control, making it difficult to adapt to the complex time-varying operating conditions of new power systems and limiting the practical application of harmonic impedance online identification technology. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention proposes a harmonic impedance identification method based on transient energy triggering and covariance correction. By employing transient energy triggering, frequency band differentiated covariance injection, and piecewise robust weight function embedding, this method effectively breaks data saturation while maintaining high RLS steady-state accuracy, enabling rapid tracking of impedance abrupt changes and online identification of broadband harmonic impedances.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, this invention proposes a harmonic impedance identification method based on transient energy triggering and covariance correction, comprising the following steps: Based on the broadband voltage and current signals obtained from the transmission line common coupling point, a harmonic impedance linear regression model is established; and based on the harmonic impedance linear regression model, a limit convergence model is constructed. Based on the wideband voltage signal, a local transient energy variation index is constructed, a comparison relationship between the local transient energy variation index and an adaptive threshold is established, and a binary state trigger signal is generated. In response to the binary state trigger signal, a frequency band differentiated dynamic covariance adaptive correction Q matrix is ​​established, wherein the injection intensity is preset according to the frequency domain response characteristics of the inverter control loop. The adaptively corrected Q matrix of the dynamic covariance is superimposed on the time update equation of the prior error covariance to modify the recursive relationship of the prior error covariance matrix, so that the prior error covariance matrix and the Kalman gain deviate from the convergence limit defined by the limit convergence model, and the piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter update. Perform matrix trace and gain limit operations on the updated harmonic impedance parameters to output broadband harmonic impedance identification results.

[0007] Furthermore, based on the broadband voltage and current signals obtained from the transmission line common coupling point, a harmonic impedance linear regression model is established. Specifically, the harmonic current observation scalar extracted based on the broadband current signal is used as the dependent variable, the harmonic voltage regression vector extracted based on the broadband voltage signal is used as the independent variable, the residual represents the model error, and the harmonic impedance to be identified is characterized by the regression parameter vector. ; in, Represents the harmonic current. Step observation scalar; Indicates the harmonic voltage number 1 Step regression vector; Indicates the harmonic impedance to be identified. Step parameter vector; Indicates the first The residual of the step.

[0008] Furthermore, based on the aforementioned harmonic impedance linear regression model, a limit convergence model is constructed, specifically as follows: Set forgetting factor Construct an iterative system for the recursive least squares method; Residual calculation: ; Kalman gain calculation: ; Parameter vector update: ; Error covariance matrix update: ; Repeat the above iterative process until the number of sampling steps... At that time, the following conditions are met: ; ; in, Indicates the first Step parameter vector estimate; Indicates the first Kalman gain of the step; Indicates the first The error covariance matrix of the step; Indicates the forgetting factor; Indicates the first Step parameter vector estimate; Represents the error covariance matrix; This represents the matrix trace operation.

[0009] Furthermore, a local transient energy variation index is constructed based on the broadband voltage signal, and a comparison relationship is established between the local transient energy variation index and an adaptive threshold to generate a binary state trigger signal; specifically: Set the length of the sliding window The harmonic voltage signal at the point of common coupling is processed using a sliding window method to calculate the average voltage within the window. : ; in, Indicates the first Instantaneous amplitude of harmonic voltage at each sampling point; Based on the average voltage within the window Calculate the local transient energy variation index : ; Set safety factor Based on historical statistical window length Calculate the adaptive threshold The local transient energy variation index at each historical moment within the historical window is marked as follows: ; ; The local transient energy variation index at the current moment With the adaptive threshold Comparison to construct binary state trigger signals:

[0010] when When, it is determined to be an impedance change condition; when The condition is determined to be either a steady-state or noise disturbance condition.

[0011] Furthermore, in response to the binary state trigger signal, a frequency band-differentiated dynamic covariance adaptive correction Q matrix is ​​established; specifically: Set the frequency band sensitivity weighting function ,in, For harmonic order; the The value is preset based on the frequency domain response characteristics of the inverter control loop; Construct the dynamic covariance adaptive correction Q matrix as follows: ; in, express An identity matrix of order 1; This represents the dimension of the parameter vector to be identified.

[0012] Furthermore, the dynamically modified covariance Q matrix is ​​superimposed onto the prior error covariance time update equation; specifically: Obtain the posterior error covariance matrix of the previous time step. ; The dynamic covariance adaptively corrected Q matrix Superimposed on the prior error covariance time update equation, the improved recursive relation of the prior error covariance matrix is ​​obtained: ; in, This represents the improved prior error covariance matrix.

[0013] Furthermore, the piecewise robust weight function adopts the IGGⅢ piecewise robust weight function, specifically: ; in, This represents the standard deviation of the residuals.

[0014] Furthermore, the piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter updates; specifically, the piecewise robust weight function is embedded into the Kalman gain equation to obtain the improved Kalman gain. Calculation formula: ; Based on the improved Kalman gain, the harmonic impedance parameters are updated: ; in, This represents the improved Kalman gain.

[0015] Furthermore, matrix trace and gain limit operations are performed on the updated harmonic impedance parameters to output broadband harmonic impedance identification results; specifically: ; in, Prior error covariance matrix; The prior error covariance matrix is ​​orthogonally corrected by the outer product operation of the gain matrix and the regression vector, and the constraint matrix always maintains positive definiteness. Calculate the trace of the posterior error covariance matrix And compare it with the convergence limit defined by the limit convergence model to verify whether the algorithm deviates from the data saturation state; Output wideband harmonic impedance identification results.

[0016] Furthermore, the identification results include the resistance component, reactance component, and susceptance component corresponding to each harmonic.

[0017] The effects described in the invention are merely those of the embodiments, and not all the effects of the invention. One of the above technical solutions has the following advantages or beneficial effects: This invention proposes a harmonic impedance identification method based on transient energy triggering and covariance correction, belonging to the technical field of harmonic impedance identification in AC distribution networks. The method includes the following steps: establishing a linear regression model of harmonic impedance based on the acquired broadband voltage and current signals at the transmission line's point of common coupling; constructing a limit convergence model based on the linear regression model; constructing a local transient energy variation index based on the broadband voltage signal, establishing a comparison relationship between the local transient energy variation index and an adaptive threshold, and generating a binary state triggering signal; responding to the binary state triggering signal... A frequency-band differentiated dynamic covariance adaptive correction Q matrix is ​​established, where the injection intensity is pre-set based on the frequency domain response characteristics of the inverter control loop. This dynamic covariance adaptive correction Q matrix is ​​superimposed onto the prior error covariance time update equation to modify the recursive relationship of the prior error covariance matrix, causing the prior error covariance matrix and Kalman gain to deviate from the convergence limit defined by the limiting convergence model. A piecewise robust weight function is embedded into the Kalman gain equation to update the harmonic impedance parameters. Matrix trace and gain limit operations are performed on the updated harmonic impedance parameters to output the broadband harmonic impedance identification result. This invention, through transient energy triggering, frequency-band differentiated covariance injection, and piecewise robust weight function embedding, effectively breaks data saturation while maintaining high forgetting factor RLS steady-state accuracy, achieving rapid tracking of impedance mutations and online identification of broadband harmonic impedances.

[0018] This invention constructs a local transient energy variation index based on a wideband voltage signal and designs a binary state trigger function. Dynamic covariance correction is initiated only when the system experiences a sudden impedance change, while covariance injection is completely disabled under steady-state conditions. This triggering mechanism requires no additional hardware, is compatible with existing power grid synchronous measurement systems, and fundamentally avoids the degradation of steady-state accuracy caused by traditional global covariance reset schemes. It achieves compatibility between high-forgetting-factor RLS steady-state high accuracy and rapid transient tracking capability.

[0019] This invention is the first to directly map the covariance injection strength to the frequency domain physical characteristics of inverter resources. Based on the inherent boundary between the PLL bandwidth and the current inner loop bandwidth, it distinguishes between the low-order harmonic accuracy-preserving frequency band and the high-order harmonic speed-tracking frequency band, configuring differentiated injection weights. The low-order harmonic frequency band is injected with extremely weak covariance, so even false triggering does not affect steady-state accuracy; the high-order harmonic frequency band is injected with strong covariance, which can instantaneously expand the covariance matrix trace by more than 106 times upon triggering, directly breaking the zero-convergence limit of RLS data saturation. This frequency band decoupling design solves the technical problem that traditional methods cannot simultaneously meet the identification requirements of the entire frequency band.

[0020] This invention rigorously demonstrates the data saturation mechanism of RLS with a high forgetting factor from a mathematical perspective. By superimposing the dynamically modified Q-matrix to the prior error covariance time update equation, the error covariance matrix deviates from the convergence limit defined by the Kalman gain limit convergence model, thus permanently breaking data saturation from a numerical perspective. Simultaneously, the closed-loop update of the posterior error covariance matrix ensures positive definiteness of the matrix, and the algorithm exhibits no divergence or accuracy decay during long-term operation.

[0021] This invention introduces an IGGⅢ piecewise robust weight function, dividing the weight function into normal, suspicious, and eliminated segments, and constructing an equivalent weight function to reduce the adverse effects of gross errors on the adjustment system. This mechanism effectively suppresses the influence of measurement noise, impulse interference, and outliers on the identification results, maintaining stable identification performance even in noisy environments.

[0022] This invention is applicable to parallel online identification of wideband harmonic channels from the 2nd to the 50th order, outputting the resistance, reactance, and susceptance components corresponding to each harmonic in real time. The entire process is based on numerical iteration and the physical characteristics of power systems, requiring no complex statistical judgments or empirical parameter adjustments. It can be directly implemented in embedded hardware and power grid online monitoring systems, demonstrating good engineering practicality and portability. This invention constructs a complete convergence verification system through matrix trace operations and gain limit operations, comparing actual identification results with the limit convergence model to determine in real time whether the algorithm deviates from data saturation, providing a quantitative basis for the reliability of the algorithm's online operation. Attached Figure Description

[0023] Figure 1A flowchart of a harmonic impedance identification method based on transient energy triggering and covariance correction is proposed for Embodiment 1 of the present invention; Figure 2 This is the convergence comparison curve of the error covariance matrix trace proposed in Embodiment 1 of the present invention; Figure 3 This is a comparison curve of the identification results under the impedance step condition proposed in Embodiment 1 of the present invention; Figure 4 This is a schematic diagram of a harmonic impedance identification system based on transient energy triggering and covariance correction proposed in Embodiment 2 of the present invention; Figure 5 This is a schematic diagram of a harmonic impedance identification device based on transient energy triggering and covariance correction proposed in Embodiment 3 of the present invention. Detailed Implementation

[0024] To clearly illustrate the technical features of this solution, the invention will be described in detail below through specific embodiments and in conjunction with the accompanying drawings. The following disclosure provides many different embodiments or examples for implementing different structures of the invention. To simplify the disclosure of the invention, components and arrangements of specific examples are described below. Furthermore, reference numerals and / or letters may be repeated in different examples. This repetition is for simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or arrangements discussed. It should be noted that the components illustrated in the drawings are not necessarily drawn to scale. Descriptions of well-known components, processing techniques, and processes are omitted in this invention to avoid unnecessarily limiting the invention.

[0025] Example 1 Embodiment 1 of this invention proposes a harmonic impedance identification method based on transient energy triggering and covariance correction, which is used to solve the technical defects of traditional high-forgetting-factor recursive least squares method in harmonic impedance identification under high-proportion inverter-type resource grid connection scenarios, such as severe data saturation and insufficient tracking ability.

[0026] This invention adopts an integrated technical solution of adaptive triggering under operating conditions, frequency band differentiated covariance injection, and robust iterative reconstruction. It clarifies the failure boundary of the algorithm through mathematical derivation, designs a non-disruptive triggering mechanism in combination with the physical characteristics of the power system, and constructs a dynamic covariance matrix that matches the frequency domain characteristics of inverter resources. Finally, it achieves steady-state high-precision identification and transient ultra-fast tracking of harmonic impedance.

[0027] Figure 1 A flowchart of a harmonic impedance identification method based on transient energy triggering and covariance correction is proposed for Embodiment 1 of the present invention; In step S1, a harmonic impedance linear regression model is established based on the obtained broadband voltage and current signals of the transmission line common coupling point; and a limit convergence model is constructed based on the harmonic impedance linear regression model. For any common connection point of a power grid transmission line, collect discrete sampling data of harmonic voltage and harmonic current output by the synchronous phasor measurement unit.

[0028] In this invention, the harmonic current observation scalar extracted based on the broadband current signal is used as the dependent variable, the harmonic voltage regression vector extracted based on the broadband voltage signal is used as the independent variable, the residual represents the model error, and the harmonic impedance to be identified is characterized by the regression parameter vector. ; in, Represents the harmonic current. Step observation scalar; Indicates the harmonic voltage number 1 Step regression vector; Indicates the harmonic impedance to be identified. Step parameter vector; Indicates the first The residual of the step.

[0029] The constructed harmonic impedance linear regression model adapted to wideband oscillation mode identification is a general standard model for power system parameter identification, which can be compatible with the independent identification of harmonics from the 2nd to the 50th order across the entire frequency band.

[0030] Based on the aforementioned harmonic impedance linear regression model, a limit convergence model is constructed, specifically as follows: Set forgetting factor Construct an iterative system for the recursive least squares method; Residual calculation: ; Kalman gain calculation: ; Parameter vector update: ; Error covariance matrix update: ; Repeat the above iterative process until the number of sampling steps... At that time, the following conditions are met: ; ; in, Indicates the first Step parameter vector estimate; Indicates the first Kalman gain of the step; Indicates the first The error covariance matrix of the step; Indicates the forgetting factor; Indicates the first Step parameter vector estimate; Represents the error covariance matrix; This represents the matrix trace operation.

[0031] This result indicates that the overall magnitude of the matrix decays infinitely, the Kalman gain, as the core weight for parameter updates, synchronously returns to zero, and the algorithm permanently loses its ability to update.

[0032] Step S1 clarifies through quantitative derivation that data saturation is an inherent mathematical property of the recursive least squares method with a high forgetting factor, which cannot be eliminated by conventional parameter adjustment. It is necessary to break the convergence limit by injecting external covariance, which is also the core theoretical basis for the design of the dynamic covariance adaptive correction Q matrix in this invention.

[0033] In step S2, a local transient energy variation index is constructed based on the wideband voltage signal, a comparison relationship is established between the local transient energy variation index and an adaptive threshold, and a binary state trigger signal is generated. This step is the core of the switching control for the dynamic covariance adaptive correction Q matrix. It aims to enable covariance injection only when impedance changes abruptly and to shut down injection throughout the steady state, thereby avoiding the damage to steady-state accuracy caused by covariance disturbances from the root cause, while also possessing extremely strong anti-interference capabilities.

[0034] The sliding window signal processing method is used to preprocess the harmonic voltage signal at the point of common coupling, and the sliding window length is set. During the preprocessing process, a recursive mean calculation is used to reduce computational complexity and avoid the redundant overhead of full summation.

[0035] The harmonic voltage signal at the point of common coupling is processed using a sliding window method, and the average voltage within the window is calculated. : ; in, Indicates the first Instantaneous amplitude of harmonic voltage at each sampling point; Based on the average voltage within the window Calculate the local transient energy variation index : ; In this step, the sliding window length The value can be set to 32, and the scope of protection of this invention is not limited to the specific lengths listed in Embodiment 1. Those skilled in the art can make reasonable choices based on the actual situation.

[0036] Local transient energy variation indicators can accurately reflect the intensity of energy fluctuations in voltage signals: when impedance changes abruptly, the voltage amplitude will undergo a step change or transient oscillation, and the sum of squares of the deviations between the voltage and the mean within the window will surge instantaneously, increasing to more than 100 times the steady-state value, forming a significant energy mutation characteristic. During steady-state operation, the energy indicator maintains a low-amplitude stable state, and voltage amplitude fluctuations are dominated by Gaussian white noise and normal load fluctuations, with fluctuation amplitudes much smaller than the rated value.

[0037] Set safety factor Based on historical statistical window length Calculate the adaptive threshold The local transient energy variation index at each historical moment within the historical window is marked as follows: ; ; In this invention, the safety factor range typically takes the following values: The selection of this safety factor range is based on the quantitative characteristics of the power grid. Provides a 20% margin to cover the maximum steady-state fluctuation of 16.7% and avoids false triggering; To ensure effective identification of the minimum 10% impedance change and avoid missed triggering, two values ​​together constitute a strictly feasible range for the trigger threshold. A safety factor of 1.5 is set here. This safety factor covers most conventional load fluctuations and Gaussian noise interference, while ensuring trigger sensitivity during actual impedance changes. The threshold is dynamically updated according to the grid operating status, automatically adapting to the grid characteristics of different voltage levels and load levels, eliminating the need for manual on-site adjustment.

[0038] The scope of protection of this invention is not limited to the specific values ​​listed in Example 1, and those skilled in the art can make reasonable selections based on the actual situation.

[0039] The local transient energy variation index at the current moment With the adaptive threshold Comparison to construct binary state trigger signals:

[0040] when When an impedance change condition is detected, indicating a real impedance change or control mode switch, dynamic Q-matrix injection is immediately initiated with a trigger delay of less than one power frequency cycle to meet the real-time requirements of transient tracking. If the condition is determined to be either steady state or noise disturbance, the dynamic covariance adaptive correction Q matrix injection is turned off, the RLS algorithm maintains the standard high forgetting factor mode, the covariance matrix is ​​not disturbed, and the steady state identification accuracy is not affected.

[0041] In step S3, in response to the binary state trigger signal, a frequency band differentiated dynamic covariance adaptive correction Q matrix is ​​established, wherein the injection intensity is preset according to the frequency domain response characteristics of the inverter control loop. Set the frequency band sensitivity weighting function ,in, For harmonic order; the The value is preset based on the frequency domain response characteristics of the inverter control loop; This step is the core innovation of this invention. Unlike traditional global injection schemes, it is the first to bind the covariance injection strength to the frequency domain physical characteristics of the inverter resource, achieving decoupling optimization between high and low frequency bands and solving the problem of global injection disrupting the steady-state accuracy of low-order harmonics. In existing technologies, fixed Q-matrix or global reset schemes indiscriminately perturb the covariance across the entire frequency band, leading to a significant increase in the steady-state error of low-order harmonics, making it impossible to balance accuracy and speed. The frequency band decoupling design of this invention configures differentiated injection strengths for the identification needs of different frequency bands, achieving dual-objective optimization of maintaining accuracy in low-order harmonics and speed tracking in high-order harmonics.

[0042] Based on the inherent bandwidth characteristics of the inverter-type resource control loop, the frequency band decoupling between the low-order harmonic steady-state convergence domain and the high-order harmonic transient response domain is realized, so that the two frequency bands are decoupled and independently controlled during the covariance correction process, fundamentally solving the contradiction that global covariance injection cannot simultaneously take into account steady-state accuracy and transient speed.

[0043] In new energy grid-connected systems, inverter-type resources adopt a dual closed-loop control structure of phase-locked loop (PLL) and current inner loop. The two types of loops have clear and fixed bandwidth boundaries: the typical engineering value of the PLL bandwidth is 50~80Hz, which mainly affects the dynamic characteristics of low-order harmonics. Within this frequency band, the system impedance is stable and fluctuates little, and the identification target is high-precision steady-state convergence; the typical engineering value of the current inner loop bandwidth is 600~800Hz, which mainly affects the dynamic characteristics of mid-to-high-order harmonics. Within this frequency band, the impedance is prone to sudden changes and saturation, and the identification target is fast parameter tracking.

[0044] Based on the system power frequency of 50Hz, the harmonic order is calculated as follows: The lower limit of the current inner loop bandwidth of 650Hz corresponds to the following harmonic order: ; therefore, This is the physical critical boundary between the frequency band dominated by the phase-locked loop and the frequency band dominated by the current inner loop. This threshold is determined by the inherent characteristics of the power electronic device control structure and has a strict engineering basis, rather than being a segmentation based on human experience.

[0045] Frequency bands are divided based on control loop characteristics: For low-sensitivity, high-precision frequency bands, For the high-sensitivity high-speed tracking frequency band, a piecewise sensitivity function is designed: ; Set as It is the injection weight of the low harmonic frequency band, and its order of magnitude is consistent with the trace of the steady-state covariance matrix. Even if the trigger signal is mistakenly set to 1, it will only cause a negligible disturbance to the covariance matrix and will not affect the steady-state identification accuracy of the low harmonics at all. Set to 0.05, this represents the injection weight for the higher harmonic frequency band, and is the difference between the trace and the steady-state covariance matrix. The magnitude of the covariance matrix trace can be instantaneously expanded upon triggering. It can be increased by more than 10 times, directly breaking the zero convergence limit of RLS data saturation and restoring the effective update capability of Kalman gain.

[0046] Generate a single-channel dynamic covariance adaptive correction Q matrix by combining the trigger signal. as follows: ; in, express An identity matrix of order 1; This represents the dimension of the parameter vector to be identified.

[0047] This matrix is ​​a positive semi-definite matrix, which ensures that the covariance is positive definite throughout the iteration process, avoids algorithm divergence, and ensures steady-state stability without disturbance and strong tracking of sudden changes.

[0048] In step S4, the dynamic covariance adaptive correction Q matrix is ​​superimposed on the prior error covariance time update equation to modify the recursive relationship of the prior error covariance matrix, so that the prior error covariance matrix and Kalman gain deviate from the convergence limit defined by the limit convergence model, and the piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter update. First, perform a priori covariance update to break the saturated core budget: The dynamic covariance adaptively corrected Q matrix Superimposed on the prior error covariance time update equation, the improved recursive relation of the prior error covariance matrix is ​​obtained: ; in, This represents the improved prior error covariance matrix.

[0049] steady state Equivalent standard algorithm; instantaneous matrix expansion when harmonic impedance undergoes a step change. Double, breaking the zero limit.

[0050] Then, an IGGⅢ robust weight function is introduced to suppress noise. The weight function is divided into a normal segment, a doubtful segment, and a rejection segment. By constructing an equivalent weight function, the adverse effects of gross errors on the adjustment system are reduced. The formula is as follows: ; in, This represents the standard deviation of the residuals.

[0051] By adding robust weight functions to suppress outliers piecewise, both robustness and accuracy are balanced.

[0052] The piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter updates; specifically, the piecewise robust weight function is embedded into the Kalman gain equation to obtain the improved Kalman gain. Calculation formula: ; In the specific calculation process, the prior covariance matrix after adaptively correcting the Q-matrix expansion using dynamic covariance is substituted into the gain equation. Robust weights are then used to reduce the weights of outliers, preventing gain distortion caused by noisy data. Under impedance abrupt change conditions, the magnitude of the prior covariance matrix increases significantly, allowing the Kalman gain to escape the zero convergence limit and recover effective parameters for weight updates. Under steady-state conditions, the gain maintains high convergence accuracy, balancing transient tracking capability with steady-state identification stability, thus completely resolving the saturation defect of traditional algorithms where the gain returns to zero from a numerical perspective.

[0053] Based on the improved Kalman gain, the harmonic impedance parameters are updated: ; in, This represents the improved Kalman gain.

[0054] In the specific implementation process, a recursive update mechanism is adopted, which can complete the calculation based solely on the parameters of the previous time step and the gain and residual of the current time step. Relying on the Kalman gain that recovers the effective amplitude, the parameters are quickly driven to converge to the true value when impedance changes abruptly, achieving millisecond-level ultra-fast tracking; during steady-state operation, the gain maintains small and stable updates, ensuring the smoothness and high accuracy of the parameter identification results, and achieving unbiased estimation of impedance parameters under all operating conditions.

[0055] Figure 2The figure shows the convergence comparison curve of the error covariance matrix trace proposed in Embodiment 1 of the present invention. The red curve in the figure represents the identification result of the improved algorithm of the present invention. It responds quickly and converges to the true value rapidly after the impedance step change, and has a faster tracking speed and a smaller overshoot. The curve comparison shows that the improved algorithm of the present invention can effectively overcome the data saturation problem of traditional RLS under the impedance change condition, has better transient tracking ability, and maintains high-precision identification performance in the steady state stage.

[0056] In step S5, matrix trace operation and gain limit operation are performed on the updated harmonic impedance parameters to output the broadband harmonic impedance identification result.

[0057] The posterior error covariance matrix is ​​updated in a closed loop. Based on the updated Kalman gain, the posterior covariance matrix is ​​corrected to provide initial matrix conditions for the next iteration, ensuring the long-term convergence stability of the algorithm. Specifically: ; in, Prior error covariance matrix; The prior error covariance matrix is ​​orthogonally corrected by the outer product operation of the gain matrix and the regression vector, ensuring that the constraint matrix remains positive definite; the trace of the posterior error covariance matrix is ​​then calculated. And compare it with the convergence limit defined by the limit convergence model to verify whether the algorithm deviates from the data saturation state; Output broadband harmonic impedance identification results. The identification results include the resistance component, reactance component, and susceptance component corresponding to each harmonic.

[0058] In the specific computation process, the prior covariance matrix is ​​orthogonally corrected by the outer product operation of the gain matrix and the regression vector, ensuring that the constraint matrix always maintains positive definiteness and avoiding iterative divergence. By constructing a complete recursive closed loop, the identification error at the current moment is fed back to the covariance matrix, achieving adaptive convergence of the error; at the same time, the optimization effect of the dynamically covariance adaptively correcting Q matrix is ​​retained, avoiding quadratic saturation caused by excessive covariance convergence and maintaining the numerical stability of the matrix, ensuring that the algorithm does not diverge or suffer from accuracy decay during long-term continuous operation.

[0059] Figure 3 The figure shows a comparison curve of identification results under the impedance step condition proposed in Embodiment 1 of the present invention; the figure contains three data sequences, namely the true value, the standard RLS identification result, and the identification result of the improved algorithm of the present invention; During the steady-state phase from t=0.00s to t=0.02s, the three data sequences remained consistent (all 2.5Ω), indicating that the improved algorithm of this invention maintained the same high-precision identification capability as the standard RLS in the steady state. At t=0.04s and after (when the impedance undergoes a step change), the true value jumps from 2.5Ω to 4.0Ω; the identification result of the standard RLS (3.0Ω) deviates significantly from the true value (4.0Ω) (error of about 1.0Ω), indicating that the traditional RLS algorithm is insufficient in tracking when the impedance changes. The identification result of the improved algorithm of this invention is completely consistent with the true value (both are 4.0Ω), realizing accurate and fast tracking of impedance step change. Figure 3 This demonstrates that the improved algorithm of this invention maintains the same high-precision identification capability as the traditional RLS in the steady-state phase, and can accurately track the real impedance value under impedance change conditions, thus overcoming the tracking failure problem caused by data saturation in the traditional RLS algorithm.

[0060] Example 2 Based on the harmonic impedance identification method based on transient energy triggering and covariance correction proposed in Embodiment 1 of this invention, Embodiment 2 of this invention also proposes a harmonic impedance identification system based on transient energy triggering and covariance correction. Figure 4 This is a schematic diagram of a harmonic impedance identification system based on transient energy triggering and covariance correction proposed in Embodiment 2 of the present invention; the system includes: The model building module is used to establish a harmonic impedance linear regression model based on the acquired broadband voltage and current signals of the transmission line common coupling point; and to construct a limit convergence model based on the harmonic impedance linear regression model. The signal generation module is used to construct a local transient energy variation index based on the wideband voltage signal, establish a comparison relationship between the local transient energy variation index and an adaptive threshold, and generate a binary state trigger signal. The matrix establishment module is used to establish a frequency band differentiated dynamic covariance adaptive correction Q matrix in response to the binary state trigger signal, wherein the injection intensity is preset according to the frequency domain response characteristics of the inverter control loop. The parameter update module is used to superimpose the dynamic covariance adaptive correction Q matrix onto the prior error covariance time update equation to modify the recursive relationship of the prior error covariance matrix, so that the prior error covariance matrix and Kalman gain deviate from the convergence limit defined by the limit convergence model, and embed the piecewise robust weight function into the Kalman gain equation to perform harmonic impedance parameter update. The calculation output module is used to perform matrix trace operation and gain limit operation on the updated harmonic impedance parameters and output the broadband harmonic impedance identification results.

[0061] The present invention provides a harmonic impedance identification system based on transient energy triggering and covariance correction, which is used to execute the harmonic impedance identification method based on transient energy triggering and covariance correction described in the present invention.

[0062] The description of the relevant parts of the harmonic impedance identification system based on transient energy triggering and covariance correction provided in Embodiment 2 of this application can be found in the detailed description of the corresponding parts in the harmonic impedance identification method based on transient energy triggering and covariance correction provided in Embodiment 1 of this application, and will not be repeated here.

[0063] Example 3 Based on the harmonic impedance identification method based on transient energy triggering and covariance correction proposed in Embodiment 1 of this invention, Embodiment 3 of this invention also proposes a harmonic impedance identification device based on transient energy triggering and covariance correction. Figure 5 This is a schematic diagram of a harmonic impedance identification device based on transient energy triggering and covariance correction proposed in Embodiment 3 of the present invention; At the hardware level, electronic device 500 includes a processor 510, and optionally, an internal bus 520, a network interface 530, and memory. The memory may include main memory, such as high-speed random-access memory (RAM), or it may also include non-volatile memory, such as at least one disk drive. Of course, the electronic device may also include other hardware required for other business operations.

[0064] The processor 510, network interface 530, and memory can be interconnected via an internal bus 520. This internal bus 520 can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be categorized as an address bus, data bus, control bus, etc. For ease of illustration, only a single bidirectional arrow is used in this diagram, but this does not imply that there is only one bus or one type of bus. The memory is used to store programs. Specifically, the program can include program code, which includes computer operation instructions. The memory can include main memory 540 and non-volatile memory 550, and provides instructions and data to the processor 510.

[0065] Processor 510 reads the corresponding computer program from non-volatile memory 550 into memory 540 and then runs it, forming a device for locating the target user at the logical level. Processor 510 executes the program stored in memory and specifically performs the following: In step S1, a harmonic impedance linear regression model is established based on the obtained broadband voltage and current signals of the transmission line common coupling point; and a limit convergence model is constructed based on the harmonic impedance linear regression model. In step S2, a local transient energy variation index is constructed based on the wideband voltage signal, a comparison relationship is established between the local transient energy variation index and an adaptive threshold, and a binary state trigger signal is generated. In step S3, in response to the binary state trigger signal, a frequency band differentiated dynamic covariance adaptive correction Q matrix is ​​established, wherein the injection intensity is preset according to the frequency domain response characteristics of the inverter control loop. In step S4, the dynamic covariance adaptive correction Q matrix is ​​superimposed on the prior error covariance time update equation to modify the recursive relationship of the prior error covariance matrix, so that the prior error covariance matrix and Kalman gain deviate from the convergence limit defined by the limit convergence model, and the piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter update. In step S5, matrix trace operation and gain limit operation are performed on the updated harmonic impedance parameters to output the broadband harmonic impedance identification result.

[0066] Figure 1It can be applied to processor 510, or implemented by processor 510. The processor may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit in the processor or by instructions in the form of software. The processor mentioned above can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. It can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in the embodiments of this application can be directly embodied as being executed by a hardware decoding processor, or executed by a combination of hardware and software modules in the decoding processor. The software module can reside in a mature storage medium in the field, such as random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, or registers. This storage medium is located in memory, and the processor reads information from the memory and, in conjunction with its hardware, completes the steps of the above method.

[0067] Of course, in addition to software implementation, the electronic device of this application does not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. In other words, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices.

[0068] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that the elements inherent in a process, method, article, or apparatus that includes a list of elements are included. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. Additionally, portions of the technical solutions provided in the embodiments of this application that are consistent with the implementation principles of corresponding technical solutions in the prior art have not been described in detail to avoid excessive elaboration.

[0069] While specific embodiments of the present invention have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of the present invention. Those skilled in the art can make other modifications or variations based on the above description. It is neither necessary nor possible to exhaustively describe all embodiments here. Various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of the present invention are still within the scope of protection of the present invention.

Claims

1. A harmonic impedance identification method based on transient energy triggering and covariance correction, characterized in that, Includes the following steps: Based on the broadband voltage and current signals obtained from the transmission line common coupling point, a harmonic impedance linear regression model is established; and based on the harmonic impedance linear regression model, a limit convergence model is constructed. Based on the wideband voltage signal, a local transient energy variation index is constructed, a comparison relationship between the local transient energy variation index and an adaptive threshold is established, and a binary state trigger signal is generated. In response to the binary state trigger signal, a frequency band differentiated dynamic covariance adaptive correction Q matrix is ​​established, wherein the injection intensity is preset according to the frequency domain response characteristics of the inverter control loop; The adaptively corrected Q matrix of the dynamic covariance is superimposed on the time update equation of the prior error covariance to modify the recursive relationship of the prior error covariance matrix, so that the prior error covariance matrix and the Kalman gain deviate from the convergence limit defined by the limit convergence model, and the piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter update. Perform matrix trace and gain limit operations on the updated harmonic impedance parameters to output broadband harmonic impedance identification results.

2. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 1, characterized in that, Based on the broadband voltage and current signals obtained from the common coupling point of the transmission line, a linear regression model of harmonic impedance is established. Specifically, the observed scalar of harmonic current extracted based on the broadband current signal is used as the dependent variable, the regression vector of harmonic voltage extracted based on the broadband voltage signal is used as the independent variable, the residual represents the model error, and the harmonic impedance to be identified is characterized by the regression parameter vector. ； in, Represents the harmonic current. Step observation scalar; Indicates the harmonic voltage number 1 Step regression vector; Indicates the harmonic impedance to be identified. Step parameter vector; Indicates the first The residual of the step.

3. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 2, characterized in that, Based on the aforementioned harmonic impedance linear regression model, a limit convergence model is constructed, specifically as follows: Set forgetting factor Construct an iterative system for the recursive least squares method; Residual calculation: ； Kalman gain calculation: ； Parameter vector update: ； Error covariance matrix update: ； Repeat the above iterative process until the number of sampling steps... At that time, the following conditions are met: ； ； in, Indicates the first Step parameter vector estimate; Indicates the first Kalman gain of the step; Indicates the first The error covariance matrix of the step; Indicates the forgetting factor; Indicates the first Step parameter vector estimate; Represents the error covariance matrix; This represents the matrix trace operation.

4. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 1, characterized in that, Based on the broadband voltage signal, a local transient energy variation index is constructed, and a comparison relationship between the local transient energy variation index and an adaptive threshold is established to generate a binary state trigger signal; specifically: Set the length of the sliding window The harmonic voltage signal at the point of common coupling is processed using a sliding window method to calculate the average voltage within the window. : ； in, Indicates the first Instantaneous amplitude of harmonic voltage at each sampling point; Based on the average voltage within the window Calculate the local transient energy variation index : ； Set safety factor Based on historical statistical window length Calculate the adaptive threshold The local transient energy variation index at each historical moment within the historical window is marked as follows: ; ； The local transient energy variation index at the current moment With the adaptive threshold Comparison to construct binary state trigger signals: when When, it is determined to be an impedance change condition; when The condition is determined to be either a steady-state or noise disturbance condition.

5. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 4, characterized in that, In response to the binary state trigger signal, an adaptively corrected Q-matrix with frequency band-differentiated dynamic covariance is established; specifically: Set the frequency band sensitivity weighting function ,in, For harmonic order; the The value is preset based on the frequency domain response characteristics of the inverter control loop; Construct the dynamic covariance adaptive correction Q matrix as follows: ； in, express An identity matrix of order 1; This represents the dimension of the parameter vector to be identified.

6. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 5, characterized in that, The dynamic covariance adaptive correction Q matrix is ​​superimposed on the prior error covariance time update equation; Specifically: Obtain the posterior error covariance matrix of the previous time step. ; The dynamic covariance adaptively corrected Q matrix Superimposed on the prior error covariance time update equation, the improved recursive relation of the prior error covariance matrix is ​​obtained: ； in, This represents the improved prior error covariance matrix.

7. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 6, characterized in that, The piecewise robust weight function adopts the IGGⅢ piecewise robust weight function, specifically: ； in, This represents the standard deviation of the residuals.

8. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 7, characterized in that, The piecewise robust weight function is embedded into the Kalman gain equation to perform harmonic impedance parameter updates; specifically, the piecewise robust weight function is embedded into the Kalman gain equation to obtain the improved Kalman gain. Calculation formula: ； Based on the improved Kalman gain, the harmonic impedance parameters are updated: ； in, This represents the improved Kalman gain.

9. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 8, characterized in that, Perform matrix trace and gain limit operations on the updated harmonic impedance parameters to output broadband harmonic impedance identification results; specifically: ； in, Prior error covariance matrix; The prior error covariance matrix is ​​orthogonally corrected by the outer product operation of the gain matrix and the regression vector, and the constraint matrix always maintains positive definiteness. Calculate the trace of the posterior error covariance matrix And compare it with the convergence limit defined by the limit convergence model to verify whether the algorithm deviates from the data saturation state; Output wideband harmonic impedance identification results.

10. The harmonic impedance identification method based on transient energy triggering and covariance correction according to claim 9, characterized in that, The identification results include the resistance component, reactance component, and susceptance component corresponding to each harmonic.

Images