A digital voltage compensation method

Abstract

The application discloses a digital voltage compensation method, and relates to the technical field of power systems, and comprises the following steps: real-time monitoring of output voltage signals and load current signals of each point of a power grid, establishment of output voltage digital sequences and load current digital sequences, extraction of dynamic characteristic variables of current load current representing change trends in future time periods, and determination of current operating state working modes; based on the voltage digital sequences, comparison with a reference voltage, calculation of real-time-reference voltage feedback compensation amounts, taking the dynamic characteristic variables as inputs, establishment of a load-voltage response prediction model, generation of voltage deviation prediction values in future control periods, and obtaining of voltage feedforward compensation amounts; according to the working mode, identification of a current working mode, giving of dynamic weight coefficients to the real-time-reference voltage feedback compensation amounts and the voltage feedforward compensation amounts, calculation of final compensation control amounts of the voltages of each point of the power grid, generation of driving signals for real-time and accurate dynamic adjustment of the voltages of each point of the power grid. The application improves the power quality of the power grid.

Description

Technical Field

[0001] This invention relates to the field of power system technology, specifically to a digital voltage compensation method. Background Technology

[0002] Current digital voltage compensation technologies suffer from the following main technical deficiencies: most solutions rely on single voltage feedback control, resulting in a delayed response to large disturbances such as load abrupt changes, and failing to effectively suppress instantaneous voltage drops or surges; although some methods introduce feedforward control, the feedforward quantity is usually based on a static load model or a fixed ratio calculation, failing to effectively utilize the dynamic trend characteristics of the load current for accurate prediction, leading to insufficient advance compensation accuracy; existing technologies lack a refined identification and adaptive control mechanism for the power grid operating state, and cannot dynamically adjust the control strategy according to different operating conditions such as small signal fluctuations, large signal transients, and steady state, resulting in an inherent contradiction between system response speed and steady-state accuracy, making it difficult to balance speed and stability; the control output and the drive signals of various actuators lack a standardized and integrated generation and adaptation mechanism, limiting the collaborative control effect and engineering applicability in complex scenarios. Summary of the Invention

[0003] To address the aforementioned technical problems, a digital voltage compensation method is provided. This technical solution solves the issues of collaborative control effectiveness and engineering applicability in the complex scenarios described above.

[0004] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0005] A digital voltage compensation method, comprising:

[0006] S1. Real-time monitoring of output voltage and load current signals at various points in the power grid, establishment of digital output voltage and load current sequences, extraction of dynamic characteristic variables representing the future trend of load current changes, and determination of the current operating status and working mode of various points in the power grid.

[0007] S2. Based on the voltage digital sequence, compare it with the reference voltage at each point of the power grid, calculate the real-time-reference voltage feedback compensation amount, and use the dynamic characteristic variable of the current load current representing the future time period change trend as input to establish a load-voltage response prediction model, generate the voltage deviation prediction value in the future control cycle, and obtain the voltage feedforward compensation amount.

[0008] S3. Integrate the real-time-reference voltage feedback compensation and the voltage feedforward compensation, identify the current working mode according to the working mode, assign dynamic weight coefficients to the real-time-reference voltage feedback compensation and the voltage feedforward compensation, calculate the final compensation control quantity of the voltage at each point of the power grid, generate drive signals, and control the actuator to perform real-time and precise dynamic adjustment of the voltage at each point of the power grid.

[0009] Preferably, step S1 specifically includes:

[0010] Based on the analog-to-digital converter, the instantaneous values ​​of output voltage and load current at each point of the power grid are collected in real time during each control cycle. Data preprocessing is performed, the length of the acquisition sliding time window is set, and the digital sequence of output voltage and load current at each point of the power grid within the time window is established.

[0011] The output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window are fitted with first-order linear least squares. The load current is used as the ordinate and the output voltage as the abscissa to obtain the fitting coefficient, which is used as the slope of the output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window.

[0012] Preferably, step S1 further includes:

[0013] Calculate the standard deviation of the digital sequence data of load current at each point in the power grid within the time window, and quantitatively assess the degree of instantaneous fluctuation of load current within the time window;

[0014] Calculate the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within adjacent time windows to obtain the slope change rate within the time window. Combine the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within the time window with the instantaneous fluctuation of the load current within the time window to extract the dynamic feature variable of the current load current that represents the trend of future changes.

[0015] Preferably, step S1 further includes:

[0016] Based on the voltage level, typical load capacity, and measurement noise level at each point in the power grid, two sets of thresholds are set: small signal and large signal. If the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window is less than the small signal threshold and the standard deviation is less than the small signal threshold, the output mode is marked as stable operating state. If the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window is greater than the large signal threshold and the standard deviation is greater than the large signal threshold, the output mode is marked as large signal transient mode. If the small signal threshold is less than the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window and less than the large signal threshold, the output mode is marked as small signal dynamic mode.

[0017] To further explain, small signals represent the magnitude of changes in background noise or minor disturbances, while large signals represent the magnitude of clear and drastic abrupt changes or shocks, and the threshold for large signals is greater than the threshold for small signals.

[0018] Based on the digital output voltage and load current sequences of each point in the power grid within the time window, and combined with the dynamic characteristic variables that represent the future trend of load current changes, the current operating status and working mode of each point in the power grid are determined.

[0019] Preferably, step S2 specifically includes:

[0020] Calculate the instantaneous voltage deviation between the output voltage at each point in the power grid and the known factory-set reference voltage within the current control cycle, and dynamically adjust the proportional gain of the current control cycle based on the absolute value of the instantaneous voltage deviation in the current control cycle.

[0021] Calculate the static error integral term at each point in the power grid within the current control cycle, prevent integral saturation based on the anti-saturation mechanism, and obtain the real-time-reference voltage feedback compensation amount for the current control cycle by combining the proportional gain of the current control cycle.

[0022] Preferably, step S2 further includes:

[0023] Based on the power grid backend storage system, historical data for L control cycles are obtained, including dynamic characteristic variables and instantaneous voltage deviations at various points in the power grid within each control cycle;

[0024] For the small-signal dynamic mode of the power grid's operation during the historical control cycle, the average load current and its digital sequence slope of each point in the power grid during each control cycle of historical length L are used as input features. For the large-signal transient mode of the power grid's operation during the historical control cycle, the instantaneous fluctuation of the load current and its rate of change of slope of each point in the power grid during each control cycle of historical length L are used as input features. The input feature parameters are initialized, and their covariance matrix is ​​calculated. For each control cycle, the latest input features are updated using recursive least squares with the actual collected instantaneous voltage deviation. The dynamic mapping relationship between the input features and the future instantaneous voltage deviation is learned, and the prediction of the instantaneous voltage deviation of the next control cycle is used as the objective to establish a load-voltage response prediction model.

[0025] Preferably, step S2 further includes:

[0026] Using the dynamic characteristic variable of the current load current representing the future trend as input, and combining the current operating mode of each point in the power grid with the average output voltage of each point in the power grid within the current control cycle as contextual auxiliary information, the load-voltage response prediction model is associated and paired, and the voltage deviation prediction value within the future control cycle is used as output to obtain the voltage feedforward compensation amount.

[0027] Preferably, step S3 specifically includes:

[0028] Based on identifying the current operating mode of each point in the power grid, different weight allocation rules are set, the fusion weight coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle is determined, and a mode-weight mapping table is established.

[0029] If in small signal dynamic mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.5 and 0.5 respectively, to solve the existing instantaneous voltage deviation and suppress continuous small fluctuations.

[0030] Preferably, step S3 further includes:

[0031] If in the large signal transient mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.2 and 0.8 respectively. Use the voltage feedforward compensation generated by the load-voltage response prediction model to quickly offset the voltage change, so as to suppress the voltage change to the greatest extent. Feedback control is used as an auxiliary to make up for the prediction error and residual deviation.

[0032] If in stable operating mode, the fusion weighting coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle are set to 0.9 and 0.1 respectively. With minimal disturbance, high-weighted feedback control is the main approach to ensure steady-state accuracy, eliminate steady-state error, and improve the response speed to future voltage changes.

[0033] Preferably, step S3 further includes:

[0034] Based on the determined weighting coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle, the final voltage compensation control quantity at each point of the power grid is calculated, and the driving signal is generated.

[0035] To further explain, the drive signals include: pulse width modulation signals: used for voltage control of inverter equipment; phase shift angle adjustment: applicable to on-load tap changer applications; and switch combination commands: used for control of static var compensator (SVG) equipment.

[0036] Based on the generated drive signal, it is sent to the IGBT, thyristor and relay control mechanisms at various points in the power grid, so that the control actuators can perform real-time and precise dynamic adjustment of the voltage at various points in the power grid.

[0037] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0038] This invention proposes a digital voltage compensation scheme. This digital voltage compensation method achieves the organic integration of closed-loop voltage feedback and feedforward control based on load trend prediction through real-time monitoring and sequence analysis. It introduces an adaptive weight allocation mechanism based on the grid operating state to dynamically optimize the control strategy under different operating conditions, thereby improving the response speed, anti-interference capability, and steady-state accuracy of voltage regulation. It realizes real-time, accurate, and adaptive dynamic compensation of grid voltage, effectively improving the power quality and operational stability of the grid. Attached Figure Description

[0039] Figure 1This is a flowchart of a digital voltage compensation method. Detailed Implementation

[0040] The following description is intended to disclose the invention and enable those skilled in the art to implement it. The preferred embodiments described below are merely examples, and other obvious variations will occur to those skilled in the art.

[0041] Reference Figure 1 As shown, a digital voltage compensation method includes:

[0042] S1. Real-time monitoring of output voltage and load current signals at various points in the power grid, establishment of digital output voltage and load current sequences, extraction of dynamic characteristic variables representing the future trend of load current changes, and determination of the current operating status and working mode of various points in the power grid.

[0043] Step S1 specifically includes:

[0044] Based on the analog-to-digital converter, the instantaneous values ​​of output voltage and load current at each point of the power grid are collected in real time during each control cycle. Data preprocessing is performed, the length of the acquisition sliding time window is set, and the digital sequence of output voltage and load current at each point of the power grid within the time window is established.

[0045] The output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window are fitted with first-order linear least squares. The load current is used as the ordinate and the output voltage as the abscissa to obtain the fitting coefficient, which is used as the slope of the output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window.

[0046] Step S1 also includes:

[0047] Calculate the standard deviation of the digital sequence data of load current at each point in the power grid within the time window, and quantitatively assess the degree of instantaneous fluctuation of load current within the time window;

[0048] Calculate the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within adjacent time windows to obtain the slope change rate within the time window. Combine the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within the time window with the instantaneous fluctuation of the load current within the time window to extract the dynamic feature variable of the current load current that represents the trend of future changes.

[0049] Step S1 also includes:

[0050] Based on the voltage level, typical load capacity, and measurement noise level at each point in the power grid, two sets of thresholds are set: small signal and large signal. If the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window is less than the small signal threshold and the standard deviation is less than the small signal threshold, the output mode is marked as stable operating state. If the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window is greater than the large signal threshold and the standard deviation is greater than the large signal threshold, the output mode is marked as large signal transient mode. If the small signal threshold is less than the absolute value of the slope of the digital output voltage sequence and the digital load current sequence at each point in the power grid within the time window and less than the large signal threshold, the output mode is marked as small signal dynamic mode.

[0051] To further explain, small signals represent the magnitude of changes in background noise or minor disturbances, while large signals represent the magnitude of clear and drastic abrupt changes or shocks, and the threshold for large signals is greater than the threshold for small signals.

[0052] Based on the digital output voltage and load current sequences of each point in the power grid within the time window, and combined with the dynamic characteristic variables that represent the future trend of load current changes, the current operating status and working mode of each point in the power grid are determined.

[0053] When using it, please refer to the steps outlined above:

[0054] In existing power grid monitoring technologies, the extraction of dynamic features from voltage and current signals typically relies on fixed threshold judgments or trend analysis of single sequences. This lacks comprehensive quantification of the real-time correlation between output characteristics and load changes, making it difficult to accurately distinguish between steady-state conditions, minor disturbances, and drastic transients. Consequently, the timeliness and adaptability of power grid operation mode identification are insufficient. This step establishes real-time digital voltage and current sequences and performs first-order linear fitting, extracting the fitting slope and its rate of change within windows. Combined with the standard deviation of load current fluctuations, dynamic feature variables are constructed. This allows for the simultaneous reflection of system output characteristics and load change trends. Based on adaptive thresholds, refined classification of operation modes is achieved, significantly improving the sensitivity of power grid condition monitoring and the accuracy of operating condition identification, providing a reliable basis for rapid response of subsequent control strategies.

[0055] S2. Based on the voltage digital sequence, compare it with the reference voltage at each point of the power grid, calculate the real-time-reference voltage feedback compensation amount, and use the dynamic characteristic variable of the current load current representing the future time period change trend as input to establish a load-voltage response prediction model, generate the voltage deviation prediction value in the future control cycle, and obtain the voltage feedforward compensation amount.

[0056] Step S2 specifically includes:

[0057] Calculate the instantaneous voltage deviation between the output voltage at each point in the power grid and the known factory-set reference voltage within the current control cycle, and dynamically adjust the proportional gain of the current control cycle based on the absolute value of the instantaneous voltage deviation in the current control cycle.

[0058] Calculate the static error integral term at each point in the power grid within the current control cycle, prevent integral saturation based on the anti-saturation mechanism, and obtain the real-time-reference voltage feedback compensation amount for the current control cycle by combining the proportional gain of the current control cycle.

[0059] Step S2 also includes:

[0060] Based on the power grid backend storage system, historical data for L control cycles are obtained, including dynamic characteristic variables and instantaneous voltage deviations at various points in the power grid within each control cycle;

[0061] For the small-signal dynamic mode of the power grid's operation during the historical control cycle, the average load current and its digital sequence slope of each point in the power grid during each control cycle of historical length L are used as input features. For the large-signal transient mode of the power grid's operation during the historical control cycle, the instantaneous fluctuation of the load current and its rate of change of slope of each point in the power grid during each control cycle of historical length L are used as input features. The input feature parameters are initialized, and their covariance matrix is ​​calculated. For each control cycle, the latest input features are updated using recursive least squares with the actual collected instantaneous voltage deviation. The dynamic mapping relationship between the input features and the future instantaneous voltage deviation is learned, and the prediction of the instantaneous voltage deviation of the next control cycle is used as the objective to establish a load-voltage response prediction model.

[0062] Step S2 also includes:

[0063] Using the dynamic characteristic variable of the current load current representing the future trend as input, and combining the current operating mode of each point in the power grid with the average output voltage of each point in the power grid within the current control cycle as contextual auxiliary information, the load-voltage response prediction model is associated and paired, and the voltage deviation prediction value within the future control cycle is used as output to obtain the voltage feedforward compensation amount.

[0064] When using it, please refer to the steps outlined above:

[0065] Existing methods largely rely on static reference voltage comparison and fixed parameter feedback, which are insufficient to cope with instantaneous voltage fluctuations caused by dynamic changes in load current. They also lack the ability to predict voltage deviations in future periods, resulting in delayed compensation response and poor disturbance rejection, especially prone to overshoot or regulation instability under conditions of large-signal transient load changes. The advantages of this approach are that by combining real-time feedback and dynamic feedforward compensation, it achieves rapid tracking and suppression of voltage deviations; by establishing a load-voltage response prediction model using dynamic characteristic variables of the load current, it improves the predictive compensation capability for future voltage deviations; by adaptively selecting input features according to different operating modes, it enhances the model's adaptability to complex operating conditions; and by updating model parameters online through recursive least squares, it improves compensation accuracy and system robustness, and reduces steady-state error and instantaneous overshoot.

[0066] S3. Integrate the real-time-reference voltage feedback compensation and the voltage feedforward compensation, identify the current working mode according to the working mode, assign dynamic weight coefficients to the real-time-reference voltage feedback compensation and the voltage feedforward compensation, calculate the final compensation control quantity of the voltage at each point of the power grid, generate drive signals, and control the actuator to perform real-time and precise dynamic adjustment of the voltage at each point of the power grid.

[0067] Step S3 specifically includes:

[0068] Based on identifying the current operating mode of each point in the power grid, different weight allocation rules are set, the fusion weight coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle is determined, and a mode-weight mapping table is established.

[0069] If in small signal dynamic mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.5 and 0.5 respectively, to solve the existing instantaneous voltage deviation and suppress continuous small fluctuations.

[0070] Step S3 also includes:

[0071] If in the large signal transient mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.2 and 0.8 respectively. Use the voltage feedforward compensation generated by the load-voltage response prediction model to quickly offset the voltage change, so as to suppress the voltage change to the greatest extent. Feedback control is used as an auxiliary to make up for the prediction error and residual deviation.

[0072] If in stable operating mode, the fusion weighting coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle are set to 0.9 and 0.1 respectively. With minimal disturbance, high-weighted feedback control is the main approach to ensure steady-state accuracy, eliminate steady-state error, and improve the response speed to future voltage changes.

[0073] Step S3 also includes:

[0074] Based on the determined weighting coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle, the final voltage compensation control quantity at each point of the power grid is calculated, and the driving signal is generated.

[0075] To further explain, the drive signals include: pulse width modulation signals: used for voltage control of inverter equipment; phase shift angle adjustment: applicable to on-load tap changer applications; and switch combination commands: used for control of static var compensator (SVG) equipment.

[0076] Based on the generated drive signal, it is sent to the IGBT, thyristor and relay control mechanisms at various points in the power grid, so that the control actuators can perform real-time and precise dynamic adjustment of the voltage at various points in the power grid.

[0077] When using it, please refer to the steps outlined above:

[0078] In the current field of dynamic voltage compensation control for power grids, traditional methods typically employ a single feedback or feedforward control strategy, which struggles to balance the dynamic response speed and steady-state accuracy of the system under various operating conditions. Especially when voltage experiences rapid changes or sustained small disturbances, the lack of adaptive adjustment of compensation weights based on the operating mode leads to response lag, overcompensation, or insufficient regulation accuracy, limiting the real-time performance and adaptability of voltage control. This step identifies the real-time operating mode of the power grid and dynamically allocates feedback and feedforward compensation weights, achieving precise adaptation to different operating conditions. In small-signal dynamic modes, balanced compensation is used to suppress continuous fluctuations; in large-signal transient modes, feedforward is emphasized to quickly smooth out sudden changes; and in stable modes, feedback is the primary method to improve steady-state accuracy. This effectively improves the overall system response speed, disturbance rejection capability, and voltage regulation accuracy, enabling real-time and precise dynamic regulation of voltage at various points in the power grid.

[0079] Taking the actual application scenario of the digital voltage compensation device developed by Chengdu Gongbaili Automation Equipment Co., Ltd. as an example, this method achieves precise management of low voltage at the end of the distribution area through real-time monitoring, pattern recognition, and dynamic compensation control. Specific embodiments are as follows:

[0080] In a remote area of ​​Sichuan, the terminal line is more than 2 kilometers long, with thin wire diameter and dispersed load. During peak electricity consumption, there is often a severe low voltage phenomenon with the voltage dropping below 190V. The digital voltage compensation device is deployed at poles No. 8 and No. 12. The system uses a built-in high-precision analog-to-digital converter to collect the output voltage and load current signals in real time and establish a digital sequence within a sliding time window.

[0081] The system calculates the fitting slope and standard deviation of the load current and output voltage. When the load current fluctuates violently and the rate of change of the slope suddenly increases, the system automatically identifies it as a large signal transient mode. At this time, the device mainly uses feedforward compensation with a weight of 0.8 to quickly predict the voltage deviation of the next cycle and generate a compensation signal. The IGBT control unit responds within 10ms to stabilize the user side voltage within the range of 220V±0.5%, effectively avoiding the voltage drop when high-power appliances such as rice cookers and heaters start up.

[0082] During daily operation, the device continuously monitors the voltage and current sequence. When the load is stable and the fluctuations are small, the system is determined to be in a stable working state or a small signal dynamic mode. At this time, feedback compensation is the main method, with a weight of 0.9 or a balance of 0.5. The voltage steady-state error is corrected in real time through proportional-integral control. The load-voltage response model trained with historical data is used for small-amplitude feedforward compensation to ensure that the voltage remains stable at the set value for a long time. The device has a built-in touch screen that can display the voltage curve, compensation status and mode indicator in real time. The operation and maintenance personnel can intuitively grasp the system's operating status through the interface, realizing unattended and maintenance-free operation.

[0083] Compared with traditional capacity expansion or V-UPQC technology, this method relies on the hardware platform of a digital voltage compensation device, including an isolation transformer, electronic voltage regulation unit, and military-grade main control system. It realizes integrated closed-loop control of monitoring, identification, and compensation. In actual cases, after installation, the user voltage curve changed from drastic fluctuations before installation, with a minimum of 180V, to a stable level between 220V and 230V. The device's own power consumption is less than 0.5%, with no mechanical wear and no reliance on fan cooling. It is suitable for complex distribution area environments with high temperature and humidity and multiple branch lines, significantly improving power supply reliability and power quality. It provides a replicable and scalable digital solution for low voltage management in distribution networks.

[0084] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention. The scope of protection claimed by the appended claims and their equivalents is defined.

Claims

1. A digital voltage compensation method, characterized in that, include: S1. Real-time monitoring of output voltage and load current signals at various points in the power grid, establishment of digital output voltage and load current sequences, extraction of dynamic characteristic variables representing the future trend of load current changes, and determination of the current operating status and working mode of various points in the power grid. S2. Based on the voltage digital sequence, compare it with the reference voltage at each point of the power grid, calculate the real-time-reference voltage feedback compensation amount, and use the dynamic characteristic variable of the current load current representing the future time period change trend as input to establish a load-voltage response prediction model, generate the voltage deviation prediction value in the future control cycle, and obtain the voltage feedforward compensation amount. S3. Integrate the real-time-reference voltage feedback compensation and the voltage feedforward compensation. Based on the identification of the current working mode of each point in the power grid, set different weight allocation rules, assign dynamic weight coefficients to the real-time-reference voltage feedback compensation and the voltage feedforward compensation, calculate the final compensation control quantity of the voltage at each point in the power grid, generate drive signals, and control the actuator to perform real-time and precise dynamic adjustment of the voltage at each point in the power grid.

2. The digital voltage compensation method according to claim 1, characterized in that, Step S1 specifically includes: Based on the analog-to-digital converter, the instantaneous values ​​of output voltage and load current at each point of the power grid are collected in real time during each control cycle. Data preprocessing is performed, the length of the acquisition sliding time window is set, and the digital sequence of output voltage and load current at each point of the power grid within the time window is established. The output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window are fitted with first-order linear least squares. The load current is used as the ordinate and the output voltage as the abscissa to obtain the fitting coefficient, which is used as the slope of the output voltage digital sequence and load current digital sequence of each point in the power grid within the same time window.

3. The digital voltage compensation method according to claim 2, characterized in that, Step S1 also includes: Calculate the standard deviation of the digital sequence data of load current at each point in the power grid within the time window, and quantitatively assess the degree of instantaneous fluctuation of load current within the time window; Calculate the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within adjacent time windows to obtain the slope change rate within the time window. Combine the slope of the digital output voltage sequence and the digital load current sequence of each point in the power grid within the time window with the instantaneous fluctuation of the load current within the time window to extract the dynamic feature variable of the current load current that represents the trend of future changes.

4. The digital voltage compensation method according to claim 3, characterized in that, Step S1 also includes: Based on the voltage level, typical load capacity, and measurement noise level at each point in the power grid, two sets of thresholds are set: small signal and large signal. If the absolute value of the slope of the output voltage digital sequence and the load current digital sequence at each point in the power grid within the time window is less than the small signal threshold and the standard deviation of the slope of the voltage digital sequence and the load current digital sequence is less than the small signal threshold, the output mode is marked as stable operating mode. If the absolute value of the slope of the output voltage digital sequence and the load current digital sequence at each point in the power grid within the time window is greater than the large signal threshold and the standard deviation of the slope of the voltage digital sequence and the load current digital sequence is greater than the large signal threshold, the output mode is marked as large signal transient mode. If the small signal threshold is less than the absolute value of the slope of the output voltage digital sequence and the load current digital sequence at each point in the power grid within the time window and less than the large signal threshold, the output mode is marked as small signal dynamic mode. Small signals represent signals of minute disturbance changes, while large signals represent signals of magnitude impact, and the threshold of large signals is greater than the threshold of small signals. Based on the digital output voltage and load current sequences of each point in the power grid within the time window, and combined with the dynamic characteristic variables that represent the future trend of load current changes, the current operating status and working mode of each point in the power grid are determined.

5. The digital voltage compensation method according to claim 4, characterized in that, Step S2 specifically includes: Calculate the instantaneous voltage deviation between the output voltage at each point in the power grid and the known factory-set reference voltage within the current control cycle, and dynamically adjust the proportional gain of the current control cycle based on the absolute value of the instantaneous voltage deviation in the current control cycle. Calculate the static error integral term at each point in the power grid within the current control cycle, prevent integral saturation based on the anti-saturation mechanism, and obtain the real-time-reference voltage feedback compensation amount for the current control cycle by combining the proportional gain of the current control cycle.

6. The digital voltage compensation method according to claim 5, characterized in that, Step S2 also includes: Based on the power grid backend storage system, historical data for L control cycles are obtained, including dynamic characteristic variables and instantaneous voltage deviations at various points in the power grid within each control cycle; For the power grid operating state of each point within the historical control cycle as a small-signal dynamic mode, the average load current of each point in the power grid and the slope of its digital load current sequence within each control cycle of L control cycles are used as input features. For the power grid operating state of each point within the historical control cycle as a large-signal transient mode, the instantaneous fluctuation degree of the load current of each point in the power grid and the rate of change of its slope within each control cycle of historical length L are used as input features. The input feature parameters are initialized, and their covariance matrix is ​​calculated. For each control cycle, the latest input features are updated using recursive least squares with the actual collected instantaneous voltage deviation. The dynamic mapping relationship between the input features and the future instantaneous voltage deviation is learned, and the prediction of the instantaneous voltage deviation of the next control cycle is used as the objective to establish a load-voltage response prediction model.

7. A digital voltage compensation method according to claim 6, characterized in that, Step S2 also includes: Using the dynamic characteristic variable of the current load current representing the future trend as input, and combining the current operating mode of each point in the power grid with the average output voltage of each point in the power grid within the current control cycle as contextual auxiliary information, the load-voltage response prediction model is associated and paired, and the voltage deviation prediction value within the future control cycle is used as output to obtain the voltage feedforward compensation amount.

8. A digital voltage compensation method according to claim 7, characterized in that, Step S3 specifically includes: Based on identifying the current operating mode of each point in the power grid, different weight allocation rules are set, the fusion weight coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle is determined, and a mode-weight mapping table is established. If in small signal dynamic mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.5 and 0.5 respectively, to solve the existing instantaneous voltage deviation and suppress continuous small fluctuations.

9. A digital voltage compensation method according to claim 8, characterized in that, Step S3 also includes: If in the large signal transient mode, set the fusion weight coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle to 0.2 and 0.8 respectively. Use the voltage feedforward compensation generated by the load-voltage response prediction model to quickly offset the voltage change, so as to suppress the voltage change to the greatest extent. Feedback control is used as an auxiliary to make up for the prediction error and residual deviation. If in stable operating mode, the fusion weighting coefficients of the real-time-reference voltage feedback compensation and the voltage feedforward compensation in the current control cycle are set to 0.9 and 0.1 respectively. With minimal disturbance, high-weighted feedback control is the main approach to ensure steady-state accuracy, eliminate steady-state error, and improve the response speed to future voltage changes.

10. A digital voltage compensation method according to claim 9, characterized in that, Step S3 also includes: Based on the determined weighting coefficient of the real-time-reference voltage feedback compensation and voltage feedforward compensation in the current control cycle, the final voltage compensation control quantity at each point of the power grid is calculated, and the driving signal is generated. The drive signals include: pulse width modulation signals: used for voltage control of inverter equipment; phase shift angle adjustment: suitable for on-load tap changers; and switch combination commands: used for control of static var compensators (SVG) equipment. Based on the generated drive signal, it is sent to the IGBT, thyristor and relay control mechanisms at various points in the power grid, so that the control actuators can perform real-time and precise dynamic adjustment of the voltage at various points in the power grid.

Images