Power system node equivalent inertia constant online estimation method and device based on adaptive Kalman filtering, equipment and medium

Abstract

The invention provides an electric power system node equivalent inertia constant online estimation method and device based on adaptive Kalman filtering, equipment and a medium, and belongs to the field of electric power systems. The method comprises the following steps: collecting and per-unit node frequency and active data, and calculating a frequency second derivative and an active first derivative; adopting ridge regression in a preset sliding window to obtain an initial estimated value of the equivalent inertia constant; carrying out normal updating and self-adaptive parameter adjustment by using self-adaptive Kalman filtering; performing jump detection based on the filtering output and triggering rapid parameter adjustment; executing steady-state locking and steady-state updating to suppress jitter when the jump does not occur; and outputting an online estimation result. According to the method, real-time and stable monitoring of the equivalent inertia constant can be realized under daily small disturbance data without manual disturbance, inertia level abrupt change can be quickly tracked, estimated value fluctuation is reduced, the requirements on hardware resources and computing power are low, and secondary side engineering deployment is facilitated.

Description

TECHNICAL FIELD

[0001] The present application belongs to the field of power systems, and relates to an online estimation method, device, equipment and medium for equivalent inertia constant of a power system node based on adaptive Kalman filtering. BACKGROUND

[0002] With new energy stations accessing the power grid through power electronic converters, the inertia level of the power system has decreased significantly, the sensitivity of the system to disturbances has increased, and the safe and stable operation of the system is facing severe challenges. In order to alleviate the stability problems caused by the decrease in the inertia of the power system, the academic and engineering circles have proposed various technologies such as virtual synchronous machines and virtual inertia control, and the virtual inertia is introduced into the power electronic converter to realize frequency support. However, due to the volatility of new energy output and the dynamic adjustment of control strategies, the inertia of the new power system presents complex characteristics such as time-varying, distributed and nonlinear, and the system inertia cannot be regarded as a fixed constant.

[0003] The current inertia monitoring of the new power system mainly has the following problems: 1) Dependence on large disturbance events: the equivalent inertia constant is identified through the dynamic response of the system after the occurrence of large disturbances such as faults or generator splitting, but the occurrence of large disturbances is low and uncontrollable; there is a frequency safety risk in artificially imposed disturbances, and in extreme cases it may cause instability.

[0004] 2) Offline estimation lag: based on the small disturbances caused by the fluctuations of new energy stations and loads in the new power system, the equivalent inertia of the system is estimated offline in a normal state. However, the virtual inertia can be dynamically adjusted according to the system demand and control strategy, which leads to insufficient online control of the inertia level of the system.

[0005] 3) Engineering constraints of data-driven methods: although data-driven methods based on the operation data of the new power system can obtain high accuracy, they usually rely on large-scale high-quality samples and feature engineering, have generalization uncertainty to scene changes (data drift), and have high requirements for computing power and time delay, which has high deployment and maintenance costs, and it is difficult to meet the lightweight, real-time and explainability requirements of the power grid side.

[0006] Therefore, it is urgent to propose an online estimation method and device for equivalent inertia constant which does not depend on large disturbances or artificial disturbances, can work continuously online under the condition of daily small disturbances and avoid offline estimation lag, at the same time, has lightweight computing power and real-time performance, has self-adaptive tracking ability to virtual inertia jump and strong explainability, in order to solve the deficiencies of the prior art in disturbance availability, timeliness and engineering deployment cost. SUMMARY

[0007] In view of the problems that the inertia monitoring in the current new power system is highly dependent on disturbance events, it is difficult to realize online monitoring, and high computing power is required, the application provides a new power system node equivalent inertia constant online estimation method and device based on adaptive Kalman filtering, which is used for online calculation of the equivalent inertia constant of the power system node to evaluate the size of the frequency support capability of the node to the power grid after the power grid is disturbed.

[0008] To solve the above technical problems, the application provides the following technical proposal: In a first aspect, the application provides a new power system node equivalent inertia constant online estimation method based on adaptive Kalman filtering, comprising: A data acquisition and derivative calculation step is constructed: real-time acquisition of frequency data and active power data of the power system node, online calculation of the second-order derivative of the frequency and the first-order derivative of the active power, and normalization processing of the data; An equivalent inertia constant initial estimation step of a sliding window ridge regression is constructed: within a sliding time window with a preset length, a dynamic linear relationship between the second-order derivative of the frequency and the first-order derivative of the active power is established by ridge regression to obtain an initial estimation value of the node equivalent inertia constant; An adaptive Kalman filter normal update step is constructed: taking the initial estimation value as the filter initial state, a state-measurement model based on the equivalent inertia constant is constructed, the process noise covariance and the measurement noise covariance are adaptively adjusted according to the statistical characteristics of the residual sequence, and the real-time estimation value of the equivalent inertia constant is obtained; An adaptive Kalman filter jump detection and rapid tracking step is constructed: based on the residual anomaly or frequency change rate criterion, the equivalent inertia constant jump event is detected, and when the jump is detected, the filter parameters are rapidly adjusted to accelerate tracking and suppress the divergence of the estimation value.

[0009] An equivalent inertia constant estimation value locking and steady-state update step is constructed: under the condition that no jump is detected, if the fluctuation of the real-time estimation value within a continuous time length is lower than a preset steady-state threshold, the locked estimation value steady value is output; thereafter, when the deviation between the real-time estimation value and the estimation value steady value exceeds a preset deviation threshold, a steady-state update based on weight is performed to correct the estimation value steady value; A result output step is constructed: the real-time estimation and the estimation value steady value after steady-state locking / steady-state update are output as the online estimation result of the equivalent inertia constant of the target node, which is used for power grid inertia monitoring and frequency support capability evaluation.

[0010] As a further technical solution of the application, the frequency and active power data of the monitored node are obtained in real time at a preset sampling period.

[0011] The initial estimation value of the equivalent inertia constant is calculated by the following formula: ; ; In the formula, is the discrete sequence of the second derivative of the frequency in the sliding window, is the transpose of is the discrete sequence of the first derivative of the active power in the sliding window, is the regularization coefficient, is the initial estimation value of the equivalent inertia constant, is a unit matrix, and the elements on the main diagonal are all 1, and the other elements are all 0.

[0012] The further technical scheme of the application is that the process noise covariance and the measurement noise covariance are adaptively adjusted according to the statistical characteristics of the residual sequence, that is, the process noise covariance matrix Q of the Kalman filter is dynamically adjusted based on the residual standard deviation in the sliding window and the observation noise covariance matrix R , and and are limited to ensure numerical stability.

[0013] The further technical scheme of the application is that the adaptive Kalman filter jump detection and rapid tracking step specifically includes: calculating the average value of the equivalent inertia constant estimation value result of the adaptive Kalman filter real-time correction in the jump detection window , judging whether it exceeds the jump detection threshold value , and if it exceeds the threshold value, executing the parameter rapid adjustment strategy to realize the rapid update of the process noise covariance matrix Q and the observation noise covariance matrix R ; when the residual standard deviation of the adaptive Kalman filter output in the jump recovery window is within the jump recovery threshold value , the and are restored to the parameter level before the rapid adjustment; the update process can be calculated by the following formula: ; ; In the formula, is the average value of the equivalent inertia constant estimation value in the jump detection window, is the jump detection window, , represent the filter parameters and ​Rapidly adjusted scaling factor.

[0014] As a further technical proposal of the present invention, the step of constructing the equivalent inertial constant estimate and locking and steady-state update specifically includes: calculating a length of The The mean of the estimated equivalent inertial constant of the adaptive Kalman filter output is calculated within each decision window. If all estimated values ​​within the window meet the tolerance Given the conditions, the stationary value of the estimated equivalent inertial constant is determined. for After the estimated stationary value is locked, if a deviation occurs between the new window mean and the current estimated stationary value, then the estimated stationary value of the equivalent inertia constant will be adjusted. Weighted adjustments are performed; the locking and updating process can be calculated using the following formula: ; In the formula, This is the stationary value of the estimated equivalent inertial constant. This represents the average value of the estimated equivalent inertia constant for the current window. To update the smoothing factor, This is the tolerance coefficient.

[0015] Secondly, the present invention provides a novel online estimation device for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering, comprising: The data acquisition and derivative calculation unit is used to acquire frequency and active power data of the measured nodes in the power system in real time, normalize them, and calculate the second derivative of frequency and the first derivative of active power online. The sliding window ridge regression equivalent inertial constant initial estimation unit is used to regress the dynamic linear relationship between the second derivative of frequency and the first derivative of active power online within a fixed sliding window, and output the initial value of the equivalent inertial constant estimate. The adaptive Kalman filter normal update unit is used to adaptively update parameters based on the Kalman filter residuals and to correct the initial value of the equivalent inertial constant online. The adaptive Kalman filter jump detection and fast tracking unit is used to detect whether the estimated value of the equivalent inertia constant in the normal update of the adaptive Kalman filter has jumped, and to rapidly adjust the parameters based on the detection results. The equivalent inertial constant initial value locking and steady-state update unit is used to regress the stationary value of the equivalent inertial constant estimate from the equivalent inertial constant estimate output by the adaptive Kalman filter, and to perform weighted correction based on the output result; The result output unit is used to display the estimated value of the equivalent inertial constant and the stationary value of the estimated value in real time.

[0016] The units are connected in data flow to perform the steps of claims 1-6.

[0017] In a third aspect, the present application provides an electronic device, comprising: at least one processor, and a memory connected with the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to perform the novel online estimation method of equivalent inertia constant of power system node based on adaptive Kalman filtering.

[0018] In a fourth aspect, the present application provides a computer readable storage medium, which stores computer instructions for enabling a processor to implement the novel online estimation method of equivalent inertia constant of power system node based on adaptive Kalman filtering when executed.

[0019] The present application has the following beneficial effects: 1. The present application obtains initial estimation of equivalent inertia constant through sliding window ridge regression, and performs residual-driven adaptive adjustment of process and measurement noise covariance in normal update of adaptive Kalman filtering, cooperates with estimation value locking and steady-state update mechanism, effectively reduces the fluctuation degree of equivalent inertia constant estimation curve, suppresses the measurement noise amplified by derivative operation, and avoids repeated fluctuations of output.

[0020] 2. The present application quickly detects whether the equivalent inertia constant estimation result of adaptive Kalman filtering output jumps through the adaptive Kalman filtering jump detection and fast tracking step, and rapidly adjusts the parameters based on this, realizes fast tracking of virtual inertia jump and suppresses estimation divergence; without applying controllable disturbance, online monitoring can be completed under daily operation data, and the real-time performance and robustness of inertia monitoring are improved.

[0021] 3. The novel online estimation device of equivalent inertia constant of power system node based on adaptive Kalman filtering is suitable for deployment on the secondary side of the station, meets the online processing needs in the sampling period of the field station, has low requirements for existing measurement and communication interface reconstruction, and is convenient for engineering landing and large-scale application. BRIEF DESCRIPTION OF DRAWINGS

[0022] Figure 1 The present application is a method flowchart.

[0023] Figure 2 The present application is a sliding window schematic diagram.

[0024] Figure 3 The present application is a device block diagram. DETAILED DESCRIPTION

[0025] In order to make the above objectives, features and advantages of the present application more clear and comprehensible, specific embodiments of the present application will be described in detail below with reference to the accompanying drawings.

[0026] In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present application. The present application, however, can be practiced in a variety of ways beyond the specific embodiments described herein without departing from the scope of the present application. It can be appreciated by those skilled in the art that the present application can be practiced without such specific details.

[0027] It should be noted that the technical terms or scientific terms used in the embodiments of the present application should be understood as the common meanings understood by those skilled in the art to which the present disclosure belongs, unless otherwise defined. The terms "first", "second", and similar terms used in the present disclosure do not indicate any order, number, or importance, but are used to distinguish different components. The terms "include", "contain", and similar terms mean that the elements or objects before the terms encompass the elements or objects listed after the terms and their equivalents, and do not exclude other elements or objects. The terms "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. It can be a fixed connection, or a detachable connection, or an integral part; it can be a mechanical connection, or an electrical connection; it can be directly connected, or indirectly connected through an intermediate medium; it can be the internal communication of two elements or the interaction relationship between two elements, unless otherwise defined. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances. "Up", "down", "left", "right", and the like are only used to represent the relative positional relationship, and when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0028] Secondly, "one embodiment" or "an embodiment" referred to herein means that a specific feature, structure, or characteristic can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments. It should be noted that the embodiments of the present application can be applied to any applicable scenario. Embodiment One

[0029] Reference Figure 1 The adaptive Kalman filter-based online estimation method for the equivalent inertia constant of a power system node provided by the first embodiment of the present application is as follows: In step 101, the frequency data and active power data of the power system node are collected in real time, the second-order derivative of the frequency and the first-order derivative of the active power are calculated online, and the data is normalized; In step 102, within a sliding time window with a preset length, a dynamic linear relationship between the second-order derivative of the frequency and the first-order derivative of the active power is established by ridge regression to obtain an initial estimated value of the equivalent inertia constant of the node; In step 103, the initial estimated value is used as the initial state of filtering, a state-measurement model based on the equivalent inertia constant is constructed, the process noise covariance and the measurement noise covariance are adaptively adjusted according to the statistical characteristics of the residual sequence, and the real-time estimated value of the equivalent inertia constant is obtained; In step 104, the equivalent inertia constant jump event is detected based on the residual anomaly or the frequency change rate criterion, and the filtering parameters are quickly adjusted to accelerate tracking and suppress the divergence of the estimated value result when the jump is detected; In step 105, under the condition that no jump is detected, if the fluctuation of the real-time estimated value is lower than the preset steady-state threshold within a continuous time length, the steady value of the estimated value is output; thereafter, when the deviation between the real-time estimated value and the steady value of the estimated value exceeds the preset deviation threshold, a steady-state update based on weight is performed to correct the steady value of the estimated value; In step 106, the real-time estimated value and the steady value of the estimated value after steady-state locking / steady-state update are output as the online estimated result of the equivalent inertia constant of the target node, which is used for power grid inertia monitoring and frequency support capability evaluation.

[0030] The present application provides an online estimation method for the equivalent inertia constant of a new power system node based on adaptive Kalman filtering, aiming at the problems of high dependence on disturbance events, difficulty in online monitoring and high requirement for computing power in the current inertia monitoring of new power systems. After obtaining the initial estimated value of the equivalent inertia constant of the node based on sliding window ridge regression, the initial estimated value is corrected online by the adaptive Kalman filtering algorithm, and the steady value of the estimated value of the equivalent inertia constant is locked to evaluate the size of the frequency support capability provided by the node to the power grid after the power grid is disturbed.

[0031] In step 101, the data collection and derivative calculation step is constructed, the frequency data and active power data of the monitored node are collected in real time, the second-order derivative of the frequency and the first-order derivative of the active power are calculated online, and the data is normalized; specifically including: The frequency and active data of the monitored node are acquired in real time at a set sampling period and normalized, then the second-order derivative of the frequency data and the first-order derivative of the active power are calculated, and the derivative calculation is as shown in the following formula: (1); (2); wherein, , are the discrete sequences of frequency and active power per unit at time k respectively, denotes the sampling time period, is the second derivative of frequency, is the first derivative of active power.

[0032] In step 102, the initial estimation of equivalent inertia constant by sliding window ridge regression is performed. In a sliding time window with a preset length, a dynamic linear relationship between the second derivative of frequency and the first derivative of active power is established by ridge regression to obtain an initial estimation of the equivalent inertia constant of the node; specifically including: The traditional synchronous generator moment of inertia and the virtual inertia provided by the power electronic converter are defined by the swing equation and can be quantified by the equivalent inertia constant. The per unit form of the swing equation can be expressed as: (3); wherein, denotes the first derivative of the rotor speed of the synchronous generator (SG), denotes the equivalent inertia constant, denotes the active power provided by the SG to the power grid, denotes the mechanical power of the synchronous generator (SG). Wherein can be decomposed into three parts: (4); wherein, is the power reference point given based on the current optimization scheduling result; is the active power adjusted by primary frequency modulation; is the active power adjusted by secondary frequency modulation.

[0033] In order to eliminate the influence of and on the calculation of the equivalent inertia constant, the formula is time-differentiated by simultaneously solving equation (3) and equation (4): (5); wherein, is the second derivative of the rotor speed of the synchronous generator (SG), , , , are the first derivatives of , , , respectively.

[0034] When the power grid is operating stably, When small disturbances such as renewable energy output and load fluctuations affect the grid frequency, the expression for the equivalent inertia constant can be written as: (6); In actual power grid operation, due to the presence of small disturbances such as new energy output and load fluctuations, the measurement data will inevitably be affected by noise. Therefore, equation (6) is modified as follows: (7); In the formula This represents the estimated value of the equivalent inertial constant; This represents the estimated value of the first derivative of active power; This represents the estimated value of the second derivative of the angular frequency; and These represent the errors in the first derivative of active power and the second derivative of angular frequency caused by measurement noise and minor disturbances, respectively.

[0035] Then, based on the ridge regression algorithm, a regularization coefficient is added. To control the degree of model regularization and avoid overfitting, as shown in equations (8)-(9): (8); (9); Mode shown and The dynamic relationship between them, i.e., the equivalent inertial constant t0 is the initial time, and t1 is the first sampling time. For the first Each sampling time; It is a discrete sequence of the second derivative of the frequency within a sliding window. for transpose, The first derivative of active power is a discrete sequence within the sliding window. The regularization coefficient is . This is the initial estimate of the equivalent inertial constant. It is an identity matrix, with all elements on its main diagonal being 1 and all other elements being 0.

[0036] To improve the speed of equivalent inertial constant estimation methods... The ability to adapt to sudden changes is achieved through the initial estimation step of the equivalent inertial constant using a sliding window ridge regression, which employs a sliding window to dynamically fit the equivalent inertial constant. A schematic diagram of the sliding window is shown below. Figure 2 As shown. The first [item] in the sliding window. The data for the second sampling window is: (10); In the formula, The sliding step size, This is the window length.

[0037] In step 103, the step of constructing the adaptive Kalman filter normal update involves using the initial estimated value as the initial state of the filter, constructing a state measurement model based on the equivalent inertia constant, and adaptively adjusting the process noise covariance and measurement noise covariance according to the statistical characteristics of the residual sequence to obtain the real-time estimated value of the equivalent inertia constant; specifically including: Kalman filtering (KF) is a recursive estimation method that calculates the estimated value of the current state given only the estimated value of the previous state and the observed value of the current state, without needing to store complete historical observation data. Based on this characteristic, real-time correction of the estimated equivalent inertial constant can be achieved. The KF algorithm uses the process noise covariance matrix to compare the previous estimated value with the current observed value. and observation noise covariance matrix To determine the confidence level of the two: The main influence is the degree to which KF depends on the observed values. This primarily affects the confidence level of the previous prediction. The classic KF setting is fixed. and To achieve real-time correction, it is difficult to balance fluctuation smoothing and real-time tracking capabilities when monitoring virtual inertia levels that are not fixed values. Therefore, the adaptive Kalman filter normal update step described in this invention proposes a residual-based adaptive parameter update mechanism to address this issue.

[0038] The residual of Kalman filtering (KF) refers to the difference between the actual observed value and the observed value calculated by KF. Due to the small sliding window and the presence of measurement noise, the observed value fluctuates significantly compared to the predicted value. The adaptive Kalman filter normal update step described above is based on the sliding update window. The residual standard deviation is dynamically adjusted to adjust the filter parameters to avoid the significant impact of occasional outliers of a single data point on parameter updates, thus making the parameter updates smoother. The standard deviation of the residuals can be calculated by equation (11). (11); (12); In the formula, For a moment The residual, The standard deviation of the residuals. This represents the scaling factor for parameter updates. This represents the learning rate based on the residuals, used to adjust the parameters. and Dependence on the degree of update. For time windows.

[0039] The process noise it represents mainly reflects the uncertainty in the state transition process and has relatively fixed physical properties. Therefore, in routine updates, it is important to... A unified update mechanism is adopted, as shown in equation (13). The measurement noise represented by this is easily affected by the external environment and sudden disturbances, exhibiting higher dynamic variability. Therefore, based on the magnitude of the residual standard deviation... The update settings are as shown in equation (14).

[0040] (13); (14); In the formula, , and , They are respectively and The moment and time The value of , , , , They are respectively and The baseline value and smoothing coefficient, To adaptively update the threshold.

[0041] By setting reasonable benchmark values ​​and smoothing coefficients... and Weighted updates are performed to avoid sudden fluctuations in filter parameters. Simultaneously, to prevent over-adjustment of filter parameters, [further measures are taken]. and Parameter setting adjustment range: (15); (16); in, , , , They are respectively and The upper and lower limits for updating.

[0042] In step 104, the adaptive Kalman filter jump detection and fast tracking step detects the equivalent inertia constant jump event based on the residual error anomaly or frequency change rate criterion, and quickly adjusts the filter parameters to accelerate tracking and suppress the divergence of the estimated value when a jump is detected; specifically including: When the system inertia level suddenly changes, the identification result of the equivalent inertia constant will also change suddenly. If only the above algorithm is relied on, a larger residual error will cause the parameters to suddenly rise, and the prediction value is relied on too much, resulting in a larger tracking delay. The adaptive Kalman filter jump detection and fast tracking step introduces a jump detection algorithm to avoid this problem. By setting a sliding jump detection window to identify the change of the equivalent inertia constant, the adaptive update strategy of and is adjusted accordingly.

[0043] (17); wherein, is the average value of the estimated value of the equivalent inertia constant in the jump detection window, is the jump detection window, is a summation variable used to traverse each time in the window, is the estimated value of the equivalent inertia constant in the smoothed estimated value of the state vector at time , is the estimated value of the equivalent inertia constant in the smoothed estimated value of the state vector at time , the average value of the filter output in the window is calculated by formula (17), and when the difference between the next time output value of the filter and the average value of the filter output in the window is greater than , such as: , the following adjustment strategy is executed, wherein is the jump detection threshold: (18); wherein , represent the scaling factors of the rapid adjustment of the filter parameters and , which are used to quickly amplify and reduce to improve the ability of the filter to track jumps. The filter parameters , of the last time are saved as , . After detecting that the system inertia level jumps, the observed values in the sliding jump recovery window are calculated by formula (19): standard deviation of the , immediately adjust , and R to , , after the state is stable, quickly restore the ability of the filter to smooth random large fluctuations. Wherein is the jump recovery threshold.

[0044] (19);

[0045] In step 105, the equivalent inertia constant estimate value locking and steady state updating step, under the condition that no jump is detected, if the fluctuation of the real-time estimate value in the continuous time length is lower than the preset steady state threshold, the estimate value steady value is output; thereafter, when the deviation of the real-time estimate value and the estimate value steady value exceeds the preset deviation threshold, the steady state updating based on weight is executed to correct the estimate value steady value; specifically including: In the actual power grid operation scheduling process, in order to facilitate reference and decision, the node inertia display result usually needs to be given an accurate estimate value. Therefore, the equivalent inertia constant estimate value steady value locking and steady state updating step proposes the following regression method: Calculate the average value of the equivalent inertia constant estimate value in the first determination window with the length of : (20); If all the estimate values in the window satisfy the tolerance coefficient Condition: (21); Wherein, H k is the equivalent inertia constant estimate value at time k, The system is considered to be in a steady state, and the estimate value steady value of the equivalent inertia constant can be determined as: (22); After the estimate value steady value is locked, if the deviation between the new window mean value and the current estimate value steady value occurs, such as formula (23), direct replacement will lead to frequent fluctuations of the estimate result. In order to avoid this problem, the equivalent inertia constant estimate value steady value locking and steady state updating step introduces a weighted correction strategy, such as formula (24), by weighted average between the historical estimate value steady value and the new window mean value, the smooth updating of the estimate result is realized, so as to balance stability and sensitivity.

[0046] (23); (24); In the formula, is the smoothing coefficient, is the updated estimated value at s+1 moment, is the new window mean value, is the estimated value at s moment.

[0047] In step 106, the construction result output step outputs the real-time estimated value and the estimated value stationary value updated by the steady-state locking / steady-state updating as the online estimation result of the equivalent inertia constant of the target node, for power grid inertia monitoring and frequency support capability evaluation; specifically including: The online estimation of the equivalent inertia constant of the target node matches the node identification, time stamp and estimation state, and labels and records the jump start and steady-state locking events, and outputs to the upper computer through the Ethernet, serial port and other communication interfaces, and caches and stores according to the period for historical playback and dispatching side analysis. Embodiment two

[0048] Referring to Figure 3 The application also provides a new online estimation device of the equivalent inertia constant of a power system node based on adaptive Kalman filtering, comprising: A data acquisition and derivative calculation unit 201 is used for acquiring the frequency data and active data of the measured node in the power system in real time, performing per-unit conversion, and calculating the frequency second-order derivative and the active power first-order derivative online; A sliding window ridge regression equivalent inertia constant initial estimation unit 202 is used for online regression of the dynamic linear relationship between the frequency second-order derivative and the active power first-order derivative in a fixed sliding window, and outputs the initial estimated value of the equivalent inertia constant; An adaptive Kalman filtering normal updating unit 203 is used for adaptive updating of parameters according to the Kalman filtering residual error, and online correction of the initial value of the equivalent inertia constant; An adaptive Kalman filtering jump detection and rapid tracking unit 204 is used for detecting whether the equivalent inertia constant estimated value of the adaptive Kalman filtering normal updating jumps, and performing rapid adjustment of parameters according to the detection result; An equivalent inertia constant initial value locking and steady-state updating unit 205 is used for regression of the equivalent inertia constant estimated value stationary value according to the equivalent inertia constant estimated value output by the adaptive Kalman filtering, and weighted correction according to the output result; A result output unit 206 is used for real-time display of the equivalent inertia constant estimated value and the estimated value stationary value.

[0049] The various variations and specific examples of the new online estimation method of the equivalent inertia constant of a power system node based on adaptive Kalman filtering in the foregoing embodiments are also applicable to the new online estimation device of the equivalent inertia constant of a power system node based on adaptive Kalman filtering in the present embodiment. Those skilled in the art can understand the new online estimation device of the equivalent inertia constant of a power system node based on adaptive Kalman filtering in the present embodiment through the foregoing detailed description of the new online estimation method of the equivalent inertia constant of a power system node based on adaptive Kalman filtering. Therefore, for the sake of brevity of the description, no further detailed description is given herein. Embodiment three

[0050] The present application provides an electronic device, comprising: at least one processor, and a memory connected to the at least one processor in communication; wherein the memory stores a computer program executable by the at least one processor, and the computer program is executed by the at least one processor to enable the at least one processor to execute the new online estimation method of the equivalent inertia constant of a power system node based on adaptive Kalman filtering.

[0051] The processor can be a central processing unit (CPU) or other forms of processing units having data processing and / or instruction execution capabilities, and can control other components in the electronic device to perform desired functions. The memory can include one or more computer program products, which can include various forms of computer readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may, for example, include random access memory (RAM), cache memory, and / or the like. The non-volatile memory may, for example, include read-only memory (ROM), hard disk, flash memory, and / or the like. One or more computer program instructions can be stored on the computer readable storage medium, and the processor can run the program instructions to implement the methods of the software programs of various embodiments of the present application described above and / or other desired functions. In addition, the electronic device can include any other appropriate components according to specific application cases. Embodiment four

[0052] The present application provides a computer readable storage medium storing computer instructions for enabling a processor to implement the new online estimation method of the equivalent inertia constant of a power system node based on adaptive Kalman filtering when executed.

[0053] In addition to the methods and apparatus described above, embodiments of the present application can also be computer program products comprising computer program instructions capable of being executed by a processor, which cause the processor to perform the steps of the methods according to various embodiments of the present application described in the above "Exemplary Methods" section of this specification. The computer program product can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, etc., and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote cloud device or server.

[0054] In addition to the methods and apparatus described above, embodiments of the present application can also be computer program products comprising computer program instructions capable of being executed by a processor, which cause the processor to perform the steps of the methods according to various embodiments of the present application described in the above "Exemplary Methods" section of this specification. The computer program product can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, etc., and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote cloud device or server.

[0055] The above description is given for illustrative and descriptive purposes. Furthermore, this description is not intended to limit embodiments of the present application to the forms disclosed herein. Although several example aspects and embodiments have been discussed above, those of ordinary skill in the art will recognize certain variations, modifications, changes, additions, and sub-combinations thereof.

Claims

1. An online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering, characterized in that, Includes the following steps: Real-time acquisition of frequency and active power data of power system nodes, online calculation of the second derivative of frequency and the first derivative of active power, and per-unit processing of the data; Within a sliding time window of a preset length, ridge regression is used to establish a dynamic linear relationship between the second derivative of frequency and the first derivative of active power, thereby obtaining an initial estimate of the nodal equivalent inertia constant. Using the initial estimated value as the initial state of filtering, a state measurement model based on the equivalent inertial constant is constructed. The process noise covariance and measurement noise covariance are adaptively adjusted according to the statistical characteristics of the residual sequence to obtain the real-time estimated value of the equivalent inertial constant. Inertia jump events are detected based on residual anomalies or frequency change rate criteria. When a jump is detected, the filter parameters are adjusted to estimate the value. If the fluctuation of the real-time estimated value is lower than the preset steady-state threshold within a continuous period of time, and no jump is detected, the steady state is locked and the stable value of the estimated value is output. When the deviation between the real-time estimated value and the stable value of the estimated value exceeds a preset deviation threshold, a weighted steady-state update is performed to correct the stable value of the estimated value. The real-time estimated value and the estimated value after steady-state locking or steady-state updating are output as the online estimation result of the equivalent inertia constant of the target node, which is used for power grid inertia monitoring and frequency support capability assessment.

2. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 1, characterized in that, The frequency and active power data of the monitored nodes are acquired in real time with a preset sampling period.

3. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 1, characterized in that, The initial estimate of the equivalent inertial constant is calculated using the following formula: ； In the formula, It is a discrete sequence of frequency second derivatives within a sliding window. for transpose, The first derivative of active power is a discrete sequence within the sliding window. The regularization coefficient is . This is the initial estimate of the equivalent inertial constant. It is an identity matrix, with all elements on its main diagonal being 1 and all other elements being 0.

4. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 1, characterized in that, The adaptive adjustment of process noise covariance and measurement noise covariance based on the statistical characteristics of the residual sequence refers to: dynamically adjusting the process noise covariance matrix of the Kalman filter based on the residual standard deviation within a sliding window. and observation noise covariance matrix and to and To ensure numerical stability, a limiting process is applied, and the update process can be calculated using the following formula: ； ； ； ； In the formula, For a moment The residual, The standard deviation of the residuals. For time window, Scaling factor For learning rate, , and , They are respectively and The moment and time The value of , , , , They are respectively and The baseline value and smoothing coefficient, To adaptively update the threshold, , , , They are respectively and The upper and lower limits for updating.

5. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 4, characterized in that, The method for detecting inertia jump events based on residual anomalies or frequency change rate criteria, and adjusting filter parameters upon detection of a jump, specifically includes: calculating the equivalent inertia constant estimate of the adaptive Kalman filter in real time within the jump detection window. The average value within the range is used to determine whether it exceeds the transition detection threshold. If the jump detection threshold is exceeded, a rapid parameter adjustment strategy is executed to achieve the process noise covariance matrix. and observation noise covariance matrix Update; when the adaptive Kalman filter output residual standard deviation is within the jump recovery window. At the jump recovery threshold Inside, then and Restore the parameter levels to their pre-rapid adjustment state; the update process can be calculated using the following formula: ； ； In the formula, This represents the average value of the estimated equivalent inertial constant within the jump detection window. For jump detection window, This is a summation variable used to iterate through each time point within the window. For at any time right time The equivalent inertia constant estimate in the smooth estimate of the state vector. , These represent the filter parameters respectively. and Rapidly adjusted scaling factor.

6. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 1, characterized in that, The steady-state locking and steady-state update process is as follows: The calculation length is... The The average value of the estimated equivalent inertial constant of the adaptive Kalman filter output is calculated within each decision window. If all estimated values ​​within the window meet the tolerance factor Under the given conditions, the stationary value of the estimated equivalent inertial constant is locked. for After the estimated stationary value is locked, if a deviation occurs between the new window mean and the current estimated stationary value, then the estimated stationary value of the equivalent inertia constant will be adjusted. Perform weighted adjustments.

7. The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering according to claim 6, characterized in that, The specific process of the weighted correction is as follows: a smoothing factor is used to perform a weighted average of the stationary value of the historical estimate and the mean of the new window.

8. A novel online estimation device for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering, characterized in that, The novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering, as described in any one of claims 1-7, includes: The data acquisition and derivative calculation unit is used to acquire frequency and active power data of the measured nodes in the power system in real time, normalize them, and calculate the second derivative of frequency and the first derivative of active power online. The sliding window ridge regression equivalent inertial constant initial estimation unit is used to regress the dynamic linear relationship between the second derivative of frequency and the first derivative of active power online within a fixed sliding window, and output the initial estimate of the equivalent inertial constant. The adaptive Kalman filter normal update unit is used to adaptively update parameters based on the Kalman filter residuals and to correct the equivalent inertia constant online. The adaptive Kalman filter jump detection and fast tracking unit is used to detect whether the estimated value of the equivalent inertia constant in the normal update of the adaptive Kalman filter has jumped, and to rapidly adjust the parameters based on the detection results. The equivalent inertial constant initial value locking and steady-state update unit is used to regress the stationary value of the equivalent inertial constant estimate from the equivalent inertial constant estimate output by the adaptive Kalman filter, and to perform weighted correction based on the output result; The result output unit is used to display the estimated value of the equivalent inertial constant and the stationary value of the estimated value in real time. The units are connected in a data stream to perform the steps described in claims 1-7.

9. An electronic device, characterized in that, The electronic device includes: At least one processor, and a memory communicatively connected to said at least one processor; The memory stores a computer program that can be executed by the at least one processor, which is then executed by the at least one processor to enable the at least one processor to perform the novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that cause a processor to execute the novel online estimation method for the equivalent inertial constant of power system nodes based on adaptive Kalman filtering, as described in any one of claims 1-7.

Images