Power distribution network voltage regulation control method and device, storage medium and electronic equipment

Abstract

The invention relates to the technical field of power distribution network voltage regulation, and discloses a power distribution network voltage regulation control method and device, a storage medium and electronic equipment, and the method comprises the steps: collecting power distribution network data through a plurality of heterogeneous data links; multi-dimensional underlying features are extracted from the power distribution network data, and the multi-dimensional underlying features comprise at least two of a root mean square amplitude, an instantaneous phase, a total harmonic distortion rate, a random run length, a nodal admittance matrix eigenvalue spectrum and a short-time waveform energy entropy; abstracting the multi-dimensional underlying features into a single-dimensional target feature, the single-dimensional target feature being a single-dimensional dynamic voltage margin; and inputting the one-dimensional target features into a preset adaptive gradient constraint electronic voltage regulation model, and solving a voltage regulation control result of the power distribution network.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of distribution network voltage regulation, and particularly relates to a distribution network voltage regulation control method and device, a storage medium and an electronic device. BACKGROUND

[0002] Under the background of the "double carbon" strategy and large-scale grid connection of distributed energy, the traditional distribution network is rapidly evolving into a highly power electronicized and actively adjustable complex system. A large number of photovoltaic inverters, flexible interconnection switches and solid-state transformers provide unprecedented voltage regulation capability at the millisecond to second level, but at the same time, they also bring the problem of voltage fluctuation and harmonic coupling across time scales. The random injection of electric vehicle charging load further amplifies the node voltage out-of-limit phenomenon, resulting in a significant decrease in power quality. The previous hierarchical decentralized voltage regulation control method relying on static models cannot balance the voltage regulation speed and coordination of multiple devices, and it is difficult to maintain voltage stability under complex working conditions. SUMMARY

[0003] The present application aims to at least solve the technical problem that the distribution network voltage regulation system in the related art cannot maintain voltage stability, has node voltage out-of-limit, and reduces power quality.

[0004] To solve the above technical problems, the embodiments of the present application provide a distribution network voltage regulation control method, comprising: acquiring distribution network data through multiple heterogeneous data links; extracting multi-dimensional bottom features from the distribution network data, wherein the multi-dimensional bottom features include at least two of root mean square amplitude, instantaneous phase, total harmonic distortion rate, random run length, node admittance matrix eigenvalue spectrum and short-time waveform energy entropy; abstracting the multi-dimensional bottom features into single-dimensional target features, wherein the single-dimensional target features are single-dimensional dynamic voltage margins; inputting the single-dimensional target features into a preset adaptive gradient constraint electronic voltage regulation model to solve a distribution network voltage regulation control result.

[0005] In some embodiments, the acquiring distribution network data through multiple heterogeneous data links comprises: acquiring three-phase voltage and current waveforms through a first μPMU arranged at the low-voltage bus side of a transformer; acquiring reactive power and DC bus voltage events transmitted by a household photovoltaic inverter at a pulse width modulation interruption through a household photovoltaic inverter data link; publishing a phasor flow through a second μPMU arranged at the end branch of the distribution network, and aligning the time with the first μPMU to acquire the harmonic current of the end branch of the distribution network; acquiring the load curve of each user through a smart meter data link.

[0006] In some embodiments, the extracting the multi-dimensional underlying features from the power grid data comprises: extracting the multi-dimensional underlying features from the power grid data according to time domain and frequency domain; The abstracting the multi-dimensional underlying features into single-dimensional target features comprises: constructing a dependency graph of the multi-dimensional underlying features by combining the maximum information coefficient method and the grey correlation degree analysis method; compressing the multi-dimensional underlying features into the single-dimensional target features by a semi-supervised Laplacian sparse dimension reduction method.

[0007] In some embodiments, the objective function of the adaptive gradient-constrained electronic voltage regulation model is: ; wherein u represents a reactive power instruction vector of all voltage regulation devices, represents a single-dimensional dynamic voltage margin, a represents a weight corresponding to a voltage allowable bandwidth, β represents a weight corresponding to a rated capacity of a voltage regulation device, and γ represents a weight corresponding to a maximum switching rate.

[0008] In some embodiments, the method further comprises: constructing optimization features, wherein the optimization features comprise an instantaneous voltage absolute fluctuation rate and a node reactive power sensitivity index; determining a lock constraint of the adaptive gradient-constrained electronic voltage regulation model according to the optimization features; optimizing the adaptive gradient-constrained electronic voltage regulation model according to the lock constraint.

[0009] In some embodiments, determining the lock constraint of the adaptive gradient-constrained electronic voltage regulation model according to the optimization features comprises: calculating a smooth sign function: ; wherein IAVR is an instantaneous voltage absolute fluctuation rate, NRSI is a node reactive power sensitivity index, and η1 and η2 are calibrated according to device physical parameters; σ is a derivable folded line function, and the output range of σ is -1-1; determining a voltage regulation mode according to the calculation result of the smooth sign function, wherein when ϕ 0, a millisecond-level fast branch is activated; when ϕ 0, a slow branch is entered; and when |ϕ| δ, the status quo is maintained.

[0010] In some embodiments, the adaptive gradient-constrained electronic voltage regulation model optimized based on the lock constraint is: .

[0011] The embodiment of the present application also provides a power distribution network voltage regulation control device, comprising: The data acquisition module is configured to acquire power distribution network data through a plurality of heterogeneous data links; The feature extraction module is configured to extract bottom-layer features from the power distribution network data and abstract the bottom-layer features into target features, wherein the target features are used to quantify voltage safety distance; The voltage regulation module is configured to input the single-dimensional target features into a preset adaptive gradient constraint electronic voltage regulation model to solve a power distribution network voltage regulation control result.

[0012] The embodiment of the present application also provides an electronic device comprising at least a memory and a processor, wherein the memory stores a computer program, and the processor implements the power distribution network voltage regulation control method described above when executing the computer program stored in the memory.

[0013] The embodiment of the present application also provides a computer readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the power distribution network voltage regulation control method described above.

[0014] The power distribution network voltage regulation control method, device, storage medium and electronic device provided by the embodiment of the present application can form an adaptive voltage regulation link from a data end to an execution end through fine-grained data acquisition, multi-dimensional target feature abstraction and adaptive gradient constraint electronic voltage regulation closed-loop control, and can fully utilize power distribution network data of different time orders to perform collaborative voltage regulation control, thereby maintaining voltage stability under complex working conditions of dynamic changes of the power distribution network, effectively reducing voltage out-of-limit rate and improving power quality. BRIEF DESCRIPTION OF DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present application or the prior art, the drawings needed in the embodiment or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments described in the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0016] Figure 1 A flowchart of the power distribution network voltage regulation control method of the embodiment of the present application; Figure 2 Another flowchart of the power distribution network voltage regulation control method of the embodiment of the present application; Figure 3Another flow chart of the power distribution network voltage regulation control method of the embodiments of the present application; Figure 4 A structural schematic diagram of the power distribution network voltage regulation control method of the embodiments of the present application. DETAILED DESCRIPTION

[0017] Various aspects and features of the present application are described herein with reference to the drawings.

[0018] It is to be understood that various alterations and modifications can be made to the embodiments of the application herein. Therefore, the above description should not be taken as limiting, but merely as exemplification of the embodiments of the application. Other modifications, being within the scope and spirit of the application, will be readily apparent to those skilled in the art.

[0019] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments of the application and, together with the general description of the application given above, and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0020] These and other characteristics of the present application will become apparent from the following description of the preferred forms given, by way of non-limiting example only, with reference to the attached drawings.

[0021] It is also to be understood that even though a number of embodiments of the application have been described herein, many modifications can be made of these embodiments without departing from the spirit and scope of the application as set forth in the following claims.

[0022] The above and other aspects, features, and advantages of the present application will become apparent from the following description of the preferred forms given, by way of non-limiting example only, with reference to the attached drawings.

[0023] Specific embodiments of the present application are described hereinafter, with reference to the accompanying drawings; however, it will be understood that the application is not limited to the specific embodiments described and shown, but encompasses many other embodiments within the spirit and scope of the application. Those skilled in the art, with appreciation of the details of the application, will be able to implement many alternative embodiments without undue experimentation.

[0024] The specification can use phrases such as "in one embodiment", "in another embodiment", "in yet another embodiment", or "in other embodiments", which can refer to one or more of the same or different embodiments of the application.

[0025] In the related art, under the condition of high penetration rate of distributed power supply (a large number of distributed power supplies are connected to the distribution network), the voltage of the low-voltage side node of the distribution network voltage regulation system frequently exceeds the limit and continuously deteriorates the power quality, which is specifically manifested in the following aspects: first, the multi-time and space scale voltage fluctuation lacks fine-grained active suppression; second, multiple types of power electronic devices such as STATCOM (Static Synchronous Compensator), OLTC (On-load tap changer, on-load tap changer or on-load tap changer) and inverter lack global cooperation, and local overshoot and circulating current frequently occur; third, the traditional voltage regulation algorithm relies too much on linearization assumption, and it is difficult to cope with the uncertainty of distributed power supply output and the significant nonlinearity of load.

[0026] Among them, STATCOM is the core device of FACTS (Flexible Alternative Current Transmission Systems), mainly used for dynamic reactive power compensation, which adjusts the reactive power in the power system through power electronic devices to improve voltage stability, suppress harmonics and improve system transient performance.

[0027] ‌OLTC is a transformer that can complete tap voltage switching in load operation, which can keep the voltage stable by adjusting the tap position of the transformer when the system voltage fluctuates or the user load changes, thereby ensuring the power supply quality, that is, it can set the voltage according to the grid load without cutting off the service.

[0028] Node voltage exceeding limit refers to the voltage of some nodes in the power system exceeding the specified range, which is usually caused by the connection of distributed photovoltaic and other renewable energy sources. The nodes in the power grid can represent various devices in the power system, such as substations, transmission lines, distribution transformers, etc.

[0029] Therefore, the embodiments of the present application provide a distribution network voltage regulation control method, device, storage medium and electronic equipment.

[0030] Embodiment one Figure 1 The flow chart of the distribution network voltage regulation control method of the embodiments of the present application is shown. As shown in Figure 1 The embodiments of the present application provide a distribution network voltage regulation control method, which comprises: S101: Collecting distribution network data through multiple heterogeneous data links.

[0031] The power distribution network data includes voltage and current waveforms, active power, reactive power, load statistical data, etc. The controller of the power system (power distribution network system) can collect fine-grained data through multiple heterogeneous data links, realize real-time collection of full-quantity information from microsecond-level synchronous waveform data to minute-level load statistical data.

[0032] S102: Extracting multi-dimensional bottom features from the power distribution network data.

[0033] After collecting the power distribution network data through different heterogeneous data links, the data is aggregated together, and then multi-dimensional bottom features are extracted therefrom. The extracted multi-dimensional bottom features can include at least two of the six categories (six dimensions) of features, i.e., root mean square voltage , instantaneous phase , total harmonic distortion , random run length , nodal admittance matrix eigenvalue spectrum , and short-time waveform energy entropy . After extracting the multi-dimensional bottom features, the multi-dimensional bottom features are abstracted and compressed into a single-dimensional target feature for quantifying the voltage safety distance. The voltage safety distance is the real-time voltage safety distance between the distributed power supply and the power distribution network.

[0034] Root mean square voltage (Root Mean Square Voltage, V rms ) refers to the effective value of periodic alternating voltage, which aims to convert the instantaneous change of alternating voltage into an equivalent direct current voltage value, and obtain the equivalent amount of alternating voltage under various waveforms and direct current voltage, so as to perform power analysis.

[0035] Instantaneous phase represents the phase shift of the signal at a given time point. Total harmonic distortion (Total Harmonic Distortion, THD) refers to the percentage of the ratio of the root mean square value of the harmonic content in the periodic alternating current to the root mean square value of the fundamental component, which is an important indicator for measuring the harmonic content in the power system. The total harmonic distortion includes harmonic voltage total distortion and / or harmonic current total distortion. The harmonic voltage total distortion is THD u = U H / U1 x 100%, where U H is the harmonic voltage content, equal to the square sum of all harmonic voltages, and U1 is the effective value of the fundamental voltage. The harmonic current total distortion THD i = I H / I1 x 100%, where I H is the harmonic current content, equal to the square sum of all harmonic currents, and I1 is the effective value of the fundamental current.

[0036] Rapid fluctuations in dynamic load current signals are called dynamic load signal run lengths. Based on the speed of fluctuation, they can be categorized into transient (1-5 power frequency cycles), short-time (5-64 power frequency cycles), and long-time (64-512 power frequency cycles) run lengths. The waveform fluctuation pattern within one run length is called the run waveform mode. According to the range switching mechanism of the Programmable Gain Amplifier (PGA) in smart meters, rapid and large fluctuations in dynamic load can cause frequent erroneous switching of the PGA gain, ultimately leading to dynamic errors in the smart meter. In this embodiment, the impact of dynamic load current run length characteristics on the dynamic error of the energy meter can be analyzed based on the acquired load statistical data, thereby improving the voltage regulation accuracy of the distribution network.

[0037] The nodal admittance matrix is ​​a matrix composed of the conductance (or resistance and admittance) values ​​between all nodes in a power grid. It is used in conjunction with the eigenvalue spectrum of the nodal admittance matrix of the power grid. It can calculate the node voltage and phase angle of each node, which facilitates subsequent voltage regulation.

[0038] Short-time waveform energy entropy It can be used to analyze short-time voltage and current waveform distortions. For example, it can be combined with load curves to analyze waveform distortions and power quality problems caused by sudden load changes, and then perform voltage regulation.

[0039] S103: Abstract the multidimensional low-level features into single-dimensional target features, wherein the single-dimensional target features are single-dimensional dynamic voltage margins.

[0040] Among them, single-dimensional target features This step is used to quantify voltage safety distance. In this step, through feature abstraction, multiple dimensions of low-level features extracted from distribution network data of different time scales are collaboratively abstracted into a single-dimensional target feature that can be understood, which facilitates subsequent distribution network voltage regulation analysis using this single-dimensional target feature.

[0041] Voltage stability margin (VSM) refers to the maximum disturbance a power system can withstand under given operating conditions without losing voltage stability. In this embodiment, dynamic voltage margin is used. As a single-dimensional target feature, it can accurately determine voltage stability, thereby enabling precise voltage regulation.

[0042] S104: Input the single-dimensional target feature into the preset adaptive gradient constraint electronic voltage regulation model to solve the voltage regulation control result of the distribution network.

[0043] The adaptive gradient constraint electronic voltage model (AGCEV) generates an execution end (control end) voltage regulation scheme according to a dynamically changing voltage safety distance through an adaptive gradient (AdaGrad) algorithm. For example, the adaptive gradient constraint electronic voltage model can improve voltage level and stability through reactive power injection (increasing reactive power) when the voltage safety distance exceeds a preset safety distance threshold, and avoid node voltage out-of-limit, thereby improving power quality. The adaptive gradient constraint electronic voltage model is an adaptive dynamic model, which can improve the voltage regulation coordination and rate of each voltage regulation device, and maintain voltage stability under complex working conditions.

[0044] The power distribution network voltage regulation control method provided by the embodiment of the application collects power distribution network data through multiple heterogeneous data links, extracts multi-dimensional bottom layer features from the power distribution network data, wherein the multi-dimensional bottom layer features include at least two of root mean square amplitude, instantaneous phase, total harmonic distortion rate, random run length, node admittance matrix eigenvalue spectrum, and short-time waveform energy entropy; the multi-dimensional bottom layer features are abstracted into single-dimensional target features, wherein the single-dimensional target features are single-dimensional dynamic voltage margins; and the single-dimensional target features are input into a preset adaptive gradient constraint electronic voltage model to solve power distribution network voltage regulation control results. The adaptive voltage regulation link from the data end to the execution end can be formed through fine-grained data collection, multi-dimensional target feature abstraction, and adaptive gradient constraint electronic voltage closed-loop control, and different time orders of power distribution network data can be fully utilized for collaborative voltage regulation control to maintain voltage stability under complex working conditions of dynamic changes of the power distribution network, effectively reduce voltage out-of-limit rate, and improve power quality.

[0045] Optionally, in step S101, the power distribution network data is collected through the constructed multiple heterogeneous data links, including: 1) three-phase voltage and current waveforms are acquired through a first μPMU arranged at a transformer low-voltage bus side; 2) reactive power and DC bus voltage events transmitted by a household photovoltaic inverter at a pulse width modulation interruption are acquired through a household photovoltaic inverter data link; 3) harmonic currents of a power distribution network terminal branch are acquired through a second μPMU arranged at the power distribution network terminal branch to publish a phasor flow and align time with the first μPMU; 4) load curves of each user are acquired through a smart meter data link.

[0046] μPMU (Micro Phasor Measurement Unit) is a phasor measurement device that uses GPS second pulse as a synchronous clock.

[0047] The first μPMU data link is formed between the first μPMU disposed at the low-voltage bus side of the transformer and the controller of the power system. The first μPMU can collect voltage and current phasor data (three-phase voltage and current waveforms) at a first preset frequency (for example, 20 kHz) and transmit them in real time to the controller of the power system through the first μPMU data link. The controller can determine whether the voltage at the low-voltage side node has exceeded the limit according to the collected three-phase voltage and current waveforms.

[0048] The control chip of the household photovoltaic inverter uploads the instantaneous reactive power and DC bus voltage events at the time of interruption of the PWM (Pulse Width Modulation) modulation signal at a preset frequency (for example, 1 kHz) through the household photovoltaic inverter data link to the controller of the power system.

[0049] The second μPMU data link is formed between the second μPMU disposed at the end branch of the distribution network and the controller of the power system. The second μPMU publishes phasor flow at 512 fps (Frames Per Second) and aligns with the GPS-PTP time to obtain harmonic current. For example, in the present embodiment, 10 kHz harmonic current can be collected.

[0050] PTP (Precision Time Protocol) and GPS are time synchronization technologies. PTP is a network time synchronization protocol based on IEEE 1588 standard. PTP aims to provide high-precision time synchronization over Ethernet, and its precision can reach sub-microsecond level. PTP synchronizes clocks between devices by sending timestamps in the network. When GPS-PTP time synchronization, the PTP module first connects to the high-precision time reference source GPS to obtain accurate time signals; PTP completes accurate time synchronization with the data collection devices (such as μPMU) of each link according to the master clock of GPS, and the time synchronization accuracy reaches microsecond or even nanosecond level.

[0051] The smart meter data link is formed between the smart meter and the controller of the power system, and the load curve and other load statistical data of each user can be transmitted back to the controller of the power system through the smart meter data link. For example, the 380V smart meter aggregates 15min load curve and power factor through NB-IoT (Narrow Band Internet of Things).

[0052] The maximum error of each link timestamp during data collection of the four data links does not exceed 100 μs, facilitating subsequent cross-scale feature fusion. In this embodiment, different heterogeneous data links can be used to collect power distribution network data of different time levels, such as microsecond-level waveform data, second-level harmonic data, minute-level load data, and hour-level topology data (such as reactive power), breaking through the limitations of traditional single time scale monitoring, and realizing holographic data perception and collection.

[0053] In some embodiments, in step S102, the multi-dimensional bottom-level features are extracted from the power distribution network data, including: S1021: Extracting multi-dimensional bottom-level features from the power distribution network data according to time domain and frequency domain.

[0054] The controller of the power system extracts multi-dimensional bottom-level features from the collected power distribution network data according to time domain and frequency domain. For example, the controller can extract features from 20 kHz waveform, 10 kHz harmonic current, 1 kHz reactive power, and 15 min load curve according to time domain and frequency domain (phasor domain), and extract the above six categories of bottom-level features.

[0055] In specific implementations, the extracted time domain features can also include maximum value, minimum value, peak-to-peak value, mean value, variance, standard deviation, mean square value, mean square error (MSE), root mean square error (RMSE), and root amplitude. Frequency domain features can also include power spectral density (Power Spectral Density), frequency distribution and peak frequency, coherence, phase response, etc.

[0056] In some embodiments, as shown in Figure 2 In step S103, the multi-dimensional bottom-level features are abstracted into single-dimensional target features, including: S1031: Constructing a dependency graph of the multi-dimensional bottom-level features by combining the maximal information coefficient method and the grey correlation degree method; S1032: Compressing the multi-dimensional bottom-level features into single-dimensional target features by semi-supervised Laplacian sparse dimension reduction.

[0057] The maximal information coefficient (MIC) method is a statistical method for measuring the strength of the relationship between two variables, including linear or nonlinear relationships. It finds an optimal discretization method and converts the calculated mutual information value into a measurement method based on this discretization method, with a value range of [0, 1]. The maximal information coefficient can effectively compensate for the defects of mutual information and more comprehensively quantify the correlation between nonlinear data.

[0058] Grey Relational Analysis (GRA) is a multi-factor statistical analysis method used to evaluate the correlation between factors in a system. The basic idea is to evaluate the correlation between factors by comparing the geometric similarity of data sequences (compare the geometric shapes of sequence curves to determine whether the relationship between different sequences is close).

[0059] In this embodiment, by combining the maximum information coefficient method and the grey correlation degree analysis method to construct a dependency graph, the correlation degree (similarity, change trend similarity) between the extracted bottom layer features can be accurately quantitatively analyzed. After constructing the dependency graph of the multi-dimensional bottom layer features, the multi-dimensional features are compressed into a single-dimensional dynamic voltage margin , thereby realizing feature abstraction. It is convenient to use the single-dimensional dynamic voltage margin for precise voltage regulation of the power distribution network.

[0060] In some embodiments, to improve the online solving speed of the AGCEV model and reduce the sensitivity to model errors, the objective function of the adaptive gradient-constrained electronic voltage regulation model is: ; where u represents the reactive power instruction vector of all voltage regulation devices, represents the single-dimensional dynamic voltage margin, a represents the weight corresponding to the voltage allowable bandwidth, b represents the weight corresponding to the rated capacity of the voltage regulation device, and g represents the weight corresponding to the maximum switching rate. argmin() represents the value of the independent variable when the given function reaches the minimum value.

[0061] The single-dimensional dynamic voltage margin is input into the above objective function to obtain the voltage regulation control result. In this embodiment, the optimal reactive power instruction is generated by constraining the modulus of the gradient and data to perform adaptive voltage regulation.

[0062] In some embodiments, as Figure 3 shown, the method further includes: S201: constructing optimization features, wherein the optimization features include instantaneous voltage absolute fluctuation rate and node reactive power sensitivity index; S202: determining the lock constraint of the adaptive gradient-constrained electronic voltage regulation model according to the optimization features; S203: optimizing the adaptive gradient-constrained electronic voltage regulation model according to the lock constraint.

[0063] In the AGCEV model, the It is still difficult to simultaneously account for millisecond-level fluctuations and second-level steady-state deviations. Therefore, in step S201, two optimization features are constructed: instantaneous voltage absolute fluctuation rate (IAVR) and node reactive power sensitivity index (NRSI), and their analysis logic, construction process, formulas, and parameter coupling relationships are separated.

[0064] The instantaneous voltage absolute fluctuation rate, by capturing the rapid amplitude of node voltage changes within 5–20 ms, is used to determine whether high-bandwidth equipment needs to quickly intervene in voltage regulation. Its specific construction process involves performing a Clarke transformation on the μPMU three-phase waveform, converting the three-phase AC signal (such as current or voltage) from the three-dimensional ABC coordinate system to the two-dimensional αβ coordinate system, thus obtaining the instantaneous amplitude in the αβ plane. , and then The formula for calculating voltage fluctuation rate, specifically the instantaneous absolute voltage fluctuation rate, is as follows: ; in, The reference voltage is the rated voltage; Δt is fixed as the fundamental frequency synchronization lock point to avoid harmonic folding.

[0065] The ability to extract high-frequency components of the molecular difference is determined by Δt. It provides a normalized scale across different voltage levels.

[0066] The larger the IAVR value, the more severe the voltage fluctuation. In the AGCEV model, the time derivative of reactive power command ∂u / ∂t and the underpenalty coefficient γ in the gradient term should be reduced simultaneously to ensure the voltage regulation response speed.

[0067] The Nodal Reactive Power Sensitivity Index (NRSI) measures the linear effect of injected reactive power on node voltage, thus determining which type of equipment should handle power factor regulation during steady-state conditions. The NRSI is constructed based on a 1-second sliding window regression, and the formula for calculating the NRSI is: ; Where Q is the reactive power sequence, V is the voltage sequence, and ε = 10⁻ 6 Used for numerical stability. and The ratio is approximately .

[0068] If NRSI increases, it indicates that the voltage boost from the same reactive power injection is more significant, and the power system will prioritize the allocation of reactive power to the inverter and suppress OLTC tap-out operations.

[0069] After constructing the optimized features in step S201, IAVR and NRSI are applied to a dual-scale gradient locking mechanism to determine the locking constraints of the AGCEV model, specifically including: S2021: calculating a smooth sign function: ; wherein IAVR is an instantaneous voltage absolute fluctuation rate, NRSI is a node reactive sensitivity index, and η1 and η2 are calibrated according to device physical parameters; σ is a derivable folded line function, and the output range of σ is -1-1; S2022: determining a voltage regulation mode according to the calculation result of the smooth sign function, wherein when ϕ 0, activating a millisecond fast branch; when ϕ 0, entering a slow branch; when | ϕ δ, maintaining the status quo.

[0070] η1 and η2 do not need artificial hyperparameters (artificially set weight tuning), and can realize voltage stability in a high penetration rate distributed power environment.

[0071] Then, a lock constraint is added to the AGCEV model improved according to the optimization characteristics, to obtain a final adaptive gradient constraint electronic voltage regulation model: .

[0072] Using the improved AGCEV model, a closed-form solution can be obtained by only one analytical iteration. In the optimized AGCEV model, the ratio of λ to α, β and γ is dynamically adjusted according to , IAVR and the current value of NRSI, so as to realize gradient adaptation.

[0073] Further, the method further comprises: S301: optimizing the adaptive gradient constraint electronic voltage regulation model according to the interaction between the instantaneous voltage absolute fluctuation rate and the node reactive sensitivity index.

[0074] There is an interaction between IAVR and NRSI: when IAVR misreports a dramatic change in voltage in the night photovoltaic scene, it will temporarily change the Q-V correlation, causing NRSI to deviate. Therefore, to avoid coupling distortion, the AGCEV model first freezes the calculation of NRSI when it detects that IAVR exceeds the preset fluctuation threshold, and then restores and updates after the fluctuation is suppressed, so as to be able to more accurately regulate voltage.

[0075] In summary, the power distribution network voltage regulation control method provided by the embodiments of the present application forms a self-adaptive voltage regulation link (full-link closed loop from raw data to control execution) from the data end to the execution end through fine-grained data collection, target feature ΔV_margin abstraction, independent construction of two-level optimization features IAVR and NRSI, and closed-loop control of double-scale gradient locking AGCEV, which does not require offline power flow matrix or manual weight tuning, can realize voltage stability and minute-level energy consumption optimization in a high penetration rate distributed power environment, maintain voltage stability compliance and suppress reactive power loss under complex working conditions; at the same time, it can realize fast power response, suppress grid harmonics, flexibly adjust reactive power, and improve power supply quality and power supply reliability. The minute-level energy consumption optimization is mainly based on the short-time 15min load curve optimization.

[0076] Embodiment Two Figure 4 The structure schematic diagram of the power distribution network voltage regulation control method of the embodiments of the present application is shown. As shown in Figure 4 The power distribution network voltage regulation control device provided by the embodiments of the present application comprises: The data collection module 10 is configured to collect power distribution network data through a plurality of heterogeneous data links; The feature extraction module 20 is configured to extract multi-dimensional bottom layer features from the power distribution network data, wherein the multi-dimensional bottom layer features include at least two of root mean square amplitude, instantaneous phase, total harmonic distortion rate, random run length, node admittance matrix eigenvalue spectrum and short-time waveform energy entropy; The feature abstraction module 30 is configured to abstract the multi-dimensional bottom layer features into single-dimensional target features, wherein the single-dimensional target features are single-dimensional dynamic voltage margin; The voltage regulation module 40 is configured to input the single-dimensional target features into a preset adaptive gradient constraint electronic voltage regulation model, and solve the power distribution network voltage regulation control result.

[0077] In some embodiments, the data collection module 10 is further configured to: Obtain three-phase voltage and current waveforms through the first μPMU arranged at the low-voltage bus side of the transformer; Obtain the reactive power and DC bus voltage events transmitted by the household photovoltaic inverter at the pulse width modulation interruption through the household photovoltaic inverter data link; Obtain the harmonic current of the power distribution network terminal branch by aligning the time of the first μPMU and the second μPMU arranged at the terminal branch of the power distribution network to release the phasor flow; Obtain the load curve of each user through the smart meter data link.

[0078] In some embodiments, the feature extraction module 20 is further configured to: extracting multi-dimensional bottom features from the power distribution network data according to time domain and frequency domain; The feature abstraction module 30 is further configured to: construct a dependency graph of the multi-dimensional bottom features by combining the maximum information coefficient method and the grey correlation degree analysis method; compress the multi-dimensional bottom features into the single-dimensional target features by a semi-supervised Laplacian sparse dimension reduction method.

[0079] In some embodiments, an objective function of the adaptive gradient-constrained electronic voltage regulation model is: ; wherein u represents a reactive power instruction vector of all voltage regulation devices, represents a single-dimensional dynamic voltage margin, a represents a weight corresponding to a voltage allowable bandwidth, β represents a weight corresponding to a rated capacity of a voltage regulation device, and γ represents a weight corresponding to a maximum switching rate.

[0080] In some embodiments, the power distribution network voltage regulation control device further comprises an optimization module configured to: construct optimization features, wherein the optimization features comprise an instantaneous voltage absolute fluctuation rate and a node reactive power sensitivity index; determine a lock constraint of the adaptive gradient-constrained electronic voltage regulation model according to the optimization features; optimize the adaptive gradient-constrained electronic voltage regulation model according to the lock constraint.

[0081] In some embodiments, the optimization module is further configured to: calculate a smooth sign function: ; wherein IAVR is the instantaneous voltage absolute fluctuation rate, NRSI is the node reactive power sensitivity index, η1 and η2 are calibrated according to device physical parameters; σ is a derivable piecewise function, and an output range of σ is -1-1; determine a voltage regulation mode according to a calculation result of the smooth sign function, wherein when ϕ 0, a millisecond-level fast branch is activated; when ϕ 0, a slow branch is entered; and when |ϕ| δ, a status quo is maintained.

[0082] In some embodiments, the adaptive gradient-constrained electronic voltage regulation model optimized based on the lock constraint is: .

[0083] The power distribution network voltage regulation control device provided by the embodiments of the present application corresponds to the power distribution network voltage regulation control method of the above embodiments, and any optional item in the power distribution network voltage regulation control method embodiments is also applicable to the embodiments of the power distribution network voltage regulation control device, which will not be described here.

[0084] Embodiment three The embodiments of the present application also provide a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the power distribution network voltage regulation control method.

[0085] The computer readable storage medium of the embodiments of the present application can adopt any combination of one or more computer readable media. The computer readable medium can be a computer readable signal medium or a computer readable storage medium. The computer readable storage medium may, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. In the embodiments of the present application, the computer readable storage medium can be any tangible medium containing or storing a program that can be used or combined with an instruction execution system, device or apparatus, for example, the above-mentioned memory.

[0086] The computer program of the embodiments of the present application can be organized into one or more computer executable components or modules. Any number and combination of such components or modules can be used to implement aspects of the present application. For example, aspects of the present application are not limited to the specific computer executable instructions or specific components or modules illustrated in the figures and described herein. Other embodiments can include different computer executable instructions or components having more or less functionality than described herein.

[0087] Embodiment four The embodiments of the present application also provide an electronic device, which at least includes a memory and a processor, the memory stores a computer program, and the processor implements the power distribution network voltage regulation control method when executing the computer program on the memory.

[0088] In some embodiments, the processor executing the computer program can be a processing device including one or more general processing devices, such as a microprocessor, a central processing unit (CPU), a graphics processing unit (GPU), and the like. More specifically, the processor can be a complex instruction set computing (CISC) microprocessor, a reduced instruction set computing (RISC) microprocessor, a very long instruction word (VLIW) microprocessor, a processor operating other instruction sets, or a processor operating a combination of instruction sets. The processor can also be one or more special-purpose processing devices such as an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), a digital signal processor (DSP), a system on a chip (SoC), and the like.

[0089] The memory can be read-only memory (ROM), random access memory (RAM), phase change random access memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), electrically erasable programmable read-only memory (EEPROM), other types of random access memory (RAM), flash memory or other forms of flash storage, cache, register, static memory, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical memory, magnetic cassettes or other magnetic storage devices, or any other possible non-transitory medium that can be used to store information or instructions that can be accessed by a computing device, etc.

[0090] The electronic device of the embodiments of the present application can include, but is not limited to, fixed terminal devices such as servers, desktop computers, digital TVs, etc., and mobile terminal devices such as in-vehicle devices (e.g., head-up display devices), handheld devices (e.g., mobile phones, tablet computers, etc.), wearable devices (e.g., smart watches, smart bands, etc.), etc.

[0091] The above description is merely preferred embodiments of the present application and a description of the principles of the technology used. Those skilled in the art should understand that the disclosed scope of the present application is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combinations of the above technical features or their equivalent features without departing from the above disclosed concept. For example, the above features are replaced with the technical features disclosed in the present application (but not limited to) having similar functions to form technical solutions.

Claims

1. A voltage regulation control method for a power distribution network, characterized in that, include: Data from the power distribution network is collected through multiple heterogeneous data links; Multidimensional low-level features are extracted from the power distribution network data, wherein the multidimensional low-level features include at least two of the following: root mean square amplitude, instantaneous phase, total harmonic distortion rate, random run length, eigenvalue spectrum of nodal admittance matrix, and short-time waveform energy entropy. The multidimensional low-level features are abstracted into single-dimensional target features, wherein the single-dimensional target feature is a single-dimensional dynamic voltage margin; The single-dimensional target feature is input into a preset adaptive gradient-constrained electronic voltage regulation model to solve for the voltage regulation control result of the distribution network.

2. The method according to claim 1, characterized in that, The process of collecting distribution network data through multiple heterogeneous data links includes: The three-phase voltage and current waveforms are obtained by sampling the first μPMU set on the low-voltage bus side of the transformer. The reactive power and DC bus voltage events transmitted by the residential photovoltaic inverter during pulse width modulation interruption are obtained through the residential photovoltaic inverter data link. The harmonic current of the distribution network terminal branch is obtained by publishing phasor current through the second μPMU set at the end branch of the distribution network and aligning the time with the first μPMU. The load curves of each user are obtained through the smart meter data link.

3. The method according to claim 1, characterized in that, The extraction of multi-dimensional low-level features from the distribution network data includes: Multidimensional low-level features are extracted from the power distribution network data based on the time domain and frequency domain. The step of abstracting the multidimensional low-level features into single-dimensional target features includes: The dependency graph of the multidimensional low-level features is constructed by combining the maximum information coefficient method and the grey relational analysis method. The multidimensional low-level features are compressed into the single-dimensional target features by using a semi-supervised Laplacian sparse dimensionality reduction method.

4. The method according to claim 1, characterized in that, The objective function of the adaptive gradient-constrained electronic voltage regulation model is: ； Where u represents the reactive power command vector of all voltage regulating devices. α represents the single-dimensional dynamic voltage margin, β represents the weight corresponding to the voltage allowable bandwidth, β represents the weight corresponding to the rated capacity of the voltage regulating equipment, and γ represents the weight corresponding to the maximum switching rate.

5. The method according to claim 1, characterized in that, The method further includes: Construct optimized features, wherein the optimized features include instantaneous voltage absolute fluctuation rate and node reactive power sensitivity index; The locking constraints of the adaptive gradient-constrained electronic voltage regulation model are determined based on the optimization characteristics. The adaptive gradient-constrained electronic voltage regulation model is optimized based on the locking constraints.

6. The method according to claim 5, characterized in that, The locking constraints of the adaptive gradient-constrained electronic voltage regulation model are determined based on the optimization characteristics, including: Calculate the sign smoothing function: ； Wherein, IAVR is the instantaneous voltage absolute fluctuation rate, NRSI is the node reactive power sensitivity index, η1 and η2 are calibrated according to the physical parameters of the equipment; σ is a differentiable piecewise linear function, and the output range of σ is –1 to 1; The voltage regulation method is determined based on the calculation result of the smoothing sign function, wherein when ϕ At 0, activate the millisecond-level fast branch; when ϕ At 0, enter the slow branch; when |ϕ| When δ is reached, the status quo is maintained.

7. The method according to claim 6, characterized in that, The adaptive gradient-constrained electronic voltage regulation model optimized based on the aforementioned locking constraints is as follows: 。 8. A voltage regulation control device for a power distribution network, characterized in that, include: The data acquisition module is configured to acquire power distribution network data through multiple heterogeneous data links; The feature extraction module is configured to extract multi-dimensional low-level features from the power distribution network data, wherein the multi-dimensional low-level features include at least two of the following: root mean square amplitude, instantaneous phase, total harmonic distortion rate, random run length, eigenvalue spectrum of nodal admittance matrix, and short-time waveform energy entropy. The feature abstraction module is configured to abstract the multi-dimensional low-level features into a single-dimensional target feature, wherein the single-dimensional target feature is a single-dimensional dynamic voltage margin. The voltage regulation module is configured to input the single-dimensional target features into a preset adaptive gradient-constrained electronic voltage regulation model to solve for the voltage regulation control results of the distribution network.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the power distribution network voltage regulation control method according to any one of claims 1 to 7.

10. An electronic device, characterized in that, It includes at least a memory and a processor, wherein the memory stores a computer program, and the processor, when executing the computer program in the memory, implements the power distribution network voltage regulation control method according to any one of claims 1 to 7.

Images