A control method and system for a distribution communication network with land-saving and coordination

Abstract

The present invention discloses a method and system for controlling a distribution communication network with provincial and local coordination, which is applied to the technical field of distribution communication network control. The method includes performing dynamic expansion estimation processing on the transient fluctuation data of a generator set, the dynamic response quantity of a load, and the state data of a reactive power compensation device to obtain fault influence domain data; constructing a provincial prediction chain model based on the fault influence domain data to obtain a prediction result of a chain reaction; performing adaptive processing of protection setting values based on an impedance trajectory on the prediction result of the chain reaction to obtain a provincial protection instruction; processing the obtained topology data and real-time operation data of a local distribution network at the city level to obtain a local system response action instruction; and in response to the provincial protection instruction, correcting the local system response action instruction, and controlling a target distribution communication network to execute a strategy matching the correction result. The method provided by the embodiment of the present invention can achieve accurate fault recovery through provincial and local coordinated distribution communication control.

Description

Technical Field

[0001] The present invention relates to the field of power distribution communication network control technology, and in particular to a provincial and local coordinated power distribution communication network control method and system. Background Art

[0002] As an important part of the power system, the distribution network undertakes the key tasks of power distribution and power supply to users. Its operational stability and reliability directly affect the normal operation of social production and life.

[0003] In existing technologies, power companies typically deploy provincial and local dispatch systems, which are used to control power distribution within the provincial and prefectural-level cities / counties, respectively. When faced with a power grid failure, it is difficult to perceive the operating status of the entire network based on local alarms and isolated analysis, and it is impossible to accurately perform fault recovery.

[0004] It can be seen that how to achieve accurate fault recovery through provincial and local coordinated power distribution communication control has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0005] The present invention provides a provincial and local coordinated power distribution communication network control method and system, so as to achieve accurate fault recovery through provincial and local coordinated power distribution communication control.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a provincial and local coordinated power distribution communication network control method, the method comprising:

[0007] Determine the transient fluctuation data of the generator set, the dynamic response of the load, and the status data of the reactive compensation equipment of the target distribution communication network when a fault occurs;

[0008] Performing dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive compensation device status data to obtain fault impact domain data;

[0009] Constructing a provincial prediction chain model using the fault impact domain data, wherein the provincial prediction chain model is configured to process the provincial prediction chain model according to a matrix decomposition method, and obtain a chain reaction prediction result output by the provincial prediction chain model based on a deep temporal network algorithm;

[0010] Performing adaptive protection setting processing based on impedance trajectory on the chain reaction prediction results to obtain provincial protection instructions;

[0011] Monte Carlo simulation algorithm is used to process the acquired prefecture-level distribution network topology data and prefecture-level real-time operation data to obtain local system response action instructions;

[0012] In response to the provincial protection instruction, correct the local system response action instruction, and control the target distribution communication network to execute a strategy matching the correction result.

[0013] As one of the preferred solutions, before determining the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data of the target distribution communication network when a fault occurs, the provincial and local coordinated control method for the distribution communication network further includes:

[0014] Determine the fault impact range;

[0015] The determination of the fault impact range includes:

[0016] Obtain the topological structure of the target distribution communication network;

[0017] Collect the load current values and voltage deviation values of several nodes in the topological structure of the target distribution communication network before and after the fault occurs;

[0018] Extract features from all the load current values and all the voltage deviations before the fault occurs to obtain a first fault feature vector;

[0019] Extract features from all the load current values and all the voltage deviations after the fault occurs to obtain a second fault feature vector;

[0020] Conduct voltage sag fluctuation analysis on the first fault feature vector and the second fault feature vector to obtain the fault impact area;

[0021] Construct an impedance matrix for the fault impact area using Kirchhoff's current law, and solve the impedance matrix of the fault impact area to obtain the fault impact range.

[0022] As one of the preferred solutions, the dynamic expansion estimation process for the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status to obtain the fault impact domain data includes:

[0023] Based on the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data, traverse the fault impact range to obtain the fault propagation path;

[0024] Use a deep learning algorithm to extract the fault propagation path to obtain the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude;

[0025] Conduct a dynamic expansion estimation process on the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude to obtain the fault impact domain data.

[0026] As one of the preferred solutions, the provincial prediction chain model is constructed based on the fault impact domain data, where the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix decomposition method and obtain the prediction result of the chain reaction output by the provincial prediction chain model based on the deep time series network algorithm, including:

[0027] Construct a fault impact domain feature matrix based on the fault impact domain data;

[0028] Use the non - negative matrix factorization method to decompose the fault impact domain feature matrix to obtain the associated feature impact vector;

[0029] Use the deep time series network algorithm to model the chain reaction of the associated feature impact vector to obtain the prediction result of the chain reaction.

[0030] As one of the preferred solutions, the adaptive processing of the protection setting value based on the impedance trajectory is performed on the prediction result of the chain reaction to obtain the provincial protection instruction, including:

[0031] Use wavelet transform to extract the prediction result of the chain reaction to obtain the voltage fluctuation sequence;

[0032] Perform dynamic mode decomposition processing on the voltage fluctuation sequence to obtain the dominant mode matrix;

[0033] Perform mapping processing on the dominant mode matrix to obtain the admittance matrix of the power grid nodes;

[0034] Use the protection setting value adaptive algorithm based on the impedance trajectory to process the admittance matrix of the power grid nodes to obtain the provincial protection instruction.

[0035] As one of the preferred solutions, the Monte Carlo simulation algorithm is used to process the obtained topological data and real - time operation data of the prefecture - level distribution network to obtain the local system response action instruction, including:

[0036] Use the Monte Carlo simulation algorithm to process the obtained topological data and real - time operation data of the prefecture - level distribution network to obtain the fault scenario database;

[0037] According to the fault scenario database, perform forward recursion processing on the transient fluctuation data of the generator set, the dynamic response amount of the load, and the state of the reactive power compensation equipment to obtain the local system response action instruction.

[0038] As one of the preferred solutions, the provincial protection instruction includes:

[0039] Fault equipment identification, power grid operation mode, power grid risk after the fault, disposal measures, and special requirements during fault disposal.

[0040] As one of the preferred solutions, before performing dynamic expansion estimation processing on the transient fluctuation data of the generator set, the dynamic response quantity of the load, and the state data of the reactive power compensation device, the provincial and local collaborative distribution communication network control method further includes:

[0041] Preprocess the transient fluctuation data of the generator set, the dynamic response quantity of the load, and the state data of the reactive power compensation device. The steps of the preprocessing include data cleaning, data integration, data transformation, and data reduction.

[0042] As one of the preferred solutions, after controlling the target distribution communication network to execute a strategy matching the correction result, the provincial and local collaborative distribution communication network control method further includes:

[0043] Verify the strategy using the cross-validation method to evaluate the accuracy of the strategy.

[0044] Another embodiment of the present invention provides a provincial and local collaborative distribution communication network control system, the system includes:

[0045] An acquisition module, configured to determine the transient fluctuation data of the generator set, the dynamic response quantity of the load, and the state data of the reactive power compensation device of the target distribution communication network when a fault occurs;

[0046] An expansion module, configured to perform dynamic expansion estimation processing on the transient fluctuation data of the generator set, the dynamic response quantity of the load, and the state data of the reactive power compensation device to obtain fault impact domain data;

[0047] A construction module, configured to construct a provincial prediction chain model with the fault impact domain data, wherein the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix decomposition method, and obtain a chain reaction prediction result output by the provincial prediction chain model based on the deep time series network algorithm;

[0048] An optimization module, configured to perform adaptive processing of protection setting values based on the impedance trajectory on the chain reaction prediction result to obtain a provincial protection instruction;

[0049] A simulation module, configured to process the obtained prefecture-level distribution network topology data and prefecture-level real-time operation data using the Monte Carlo simulation algorithm to obtain a local system response action instruction;

[0050] An execution module, configured to respond to the provincial protection instruction, correct the local system response action instruction, and control the target distribution communication network to execute a strategy matching the correction result.

[0051] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0052] The present invention determines the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment state data of the target distribution communication network when a fault occurs; performs dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment state data to obtain fault influence domain data; constructs a provincial prediction cascade model with the fault influence domain data, wherein the provincial prediction cascade model is configured to process the provincial prediction cascade model according to the matrix decomposition method and obtain a cascade reaction prediction result output by the provincial prediction cascade model based on the deep time series network algorithm; performs adaptive processing of protection setting values based on the impedance trajectory on the cascade reaction prediction result to obtain a provincial protection instruction; uses the Monte Carlo simulation algorithm to process the obtained prefecture-level distribution network topology data and prefecture-level real-time operation data to obtain a local system response action instruction; in response to the provincial protection instruction, corrects the local system response action instruction, and controls the target distribution communication network to execute a strategy matching the correction result.

[0053] Compared with the prior art, the present invention captures the fluctuation data and state when a fault occurs, quantifies the fault influence range, and provides input data in the spatial dimension for subsequent cascade reaction prediction at the provincial level; then reduces the dimensionality of high-dimensional power grid data through the matrix decomposition method, extracts key features, identifies key nodes in the provincial power grid that are prone to trigger cascade faults, and then mines the time series features of the post-fault data through the deep time series network algorithm to predict the cascade reaction that may be triggered by the fault, anticipates the spread path and potential consequences of the fault in advance at the provincial level, and collaboratively adjusts the protection devices within the scope of the provincial power grid according to the predicted cascade reaction risk to obtain a provincial protection instruction; the Monte Carlo simulation can generate a large number of possible operation scenarios through random sampling, generate targeted local system response action instructions according to the specific topology and real-time state of the prefecture-level power grid, correct the local system response action instruction in response to the provincial protection instruction, control the target distribution communication network to execute a strategy matching the correction result, and generate a flexible response strategy based on the local instruction corrected under the provincial instruction to ensure the compatibility of local control and global stability; through this processing flow, the present invention can improve the fault self-healing ability of the distribution communication network through provincial and local collaboration and ensure the safe operation of the power system. Brief Description of the Drawings

[0054] Figure 1 is a schematic flowchart of a provincial and local collaborative control method for a distribution communication network in one embodiment of the present invention;

[0055] Figure 2 is a schematic structural diagram of a provincial and local collaborative control system for a distribution communication network in one embodiment of the present invention;

[0056] Reference Signs:

[0057] Among them, 11 is an acquisition module; 12 is an expansion module; 13 is a construction module; 14 is an optimization module; 15 is a simulation module; 16 is an execution module. Detailed implementation manners

[0058] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0059] In the description of the present invention, the terms "first", "second", "third", etc. are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, features defined with "first", "second", "third", etc. may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise stated, the meaning of "a plurality" is two or more.

[0060] In the description of the present invention, it should be noted that unless otherwise clearly defined and limited, the terms "installed", "connected", "connected" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be a mechanical connection or an electrical connection; it may be directly connected, or indirectly connected through an intermediate medium, and it may be the internal communication of two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are only for the purpose of illustration, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be understood as a limitation of the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0061] In the description of the present invention, it should be noted that unless otherwise defined, all technical and scientific terms used in the present invention have the same meaning as commonly understood by those skilled in the technical field to which the present invention belongs. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments, and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0062] As an important part of the power system, the distribution network undertakes the crucial tasks of power distribution and user power supply. Its operation stability and reliability directly affect the normal operation of social production and life. In the existing technology, power companies usually deploy provincial dispatching and local dispatching, which are respectively used for distribution control within the provincial scope and within the scope of prefecture-level cities / counties. When facing power grid faults, it is difficult to perceive the operation status of the whole network based on the local alarm and isolated analysis method, and it is impossible to accurately recover from the faults.

[0063] Therefore, how to achieve accurate fault recovery through provincial and local collaborative distribution communication control has become a technical problem that needs to be solved urgently by those skilled in the art.

[0064] To solve this problem, an embodiment of the present invention provides a method for controlling a provincial and local collaborative distribution communication network. Specifically, please refer to Figure 1 , Figure 1 which shows a schematic flow chart of the method for controlling a provincial and local collaborative distribution communication network in one embodiment of the present invention. The method includes:

[0065] S1: Determine the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data of the target distribution communication network when a fault occurs;

[0066] S2: Perform dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data to obtain fault influence domain data;

[0067] S3: Construct a provincial prediction cascading model with the fault influence domain data. Among them, the provincial prediction cascading model is configured to process the provincial prediction cascading model according to the matrix decomposition method, and obtain the cascading reaction prediction result output by the provincial prediction cascading model based on the deep time series network algorithm;

[0068] S4: Perform adaptive processing of protection setting values based on the impedance trajectory on the cascading reaction prediction result to obtain a provincial protection instruction;

[0069] S5: Use the Monte Carlo simulation algorithm to process the obtained topological data and real-time operation data of the prefecture-level distribution network to obtain a local system response action instruction;

[0070] S6: In response to the provincial protection instruction, correct the local system response action instruction, and control the target distribution communication network to execute a strategy that matches the correction result.

[0071] Specifically, before determining the key data when a fault occurs, determining the fault influence range is to improve the pertinence of data collection, calculation efficiency, and the accuracy of control strategies, and avoid resource waste and decision-making lag caused by blindly collecting and analyzing the data of the whole network.

[0072] The direct impact scope of most distribution faults is limited and usually confined to a certain feeder or the power supply area of a substation. By first determining the impact scope, data can be collected only for the nodes and equipment in the fault-related area. For example: only monitor the transient fluctuations of the generator sets within the fault-impacted area; only obtain the dynamic response quantities of the loads within the affected area; focus on the status of the reactive power compensation equipment within the affected area.

[0073] Before performing step S1, that is, before determining the transient fluctuation data of the generator sets, the dynamic response quantities of the loads, and the status data of the reactive power compensation equipment of the target distribution communication network when the fault occurs, it is necessary to first determine the impact scope of the fault, specifically including: obtaining the topological structure of the target distribution communication network; collecting the load current values and voltage deviation values of several nodes in the topological structure of the target distribution communication network before and after the fault occurs; performing feature extraction on all the load current values and all the voltage deviations before the fault occurs to obtain a first fault feature vector; performing feature extraction on all the load current values and all the voltage deviations after the fault occurs to obtain a second fault feature vector; performing voltage sag fluctuation analysis on the first fault feature vector and the second fault feature vector to obtain the fault-impacted area; constructing an impedance matrix of the fault-impacted area using Kirchhoff's current law and solving the impedance matrix of the fault-impacted area to obtain the fault impact scope.

[0074] Specifically, the topological structure is the physical carrier for fault analysis. Only by clarifying the network architecture of the power grid can the fault location be located and the fault propagation path be deduced. Subsequent current / voltage analysis and impedance matrix construction are both based on the topological structure. For example, the electrical connection characteristics of the power grid are described through the node admittance matrix or impedance matrix.

[0075] Collect the load current values and voltage deviation values of several nodes in the topological structure of the target distribution communication network before and after the fault occurs. Among them, the load current value reflects the active / reactive power flow situation of each node before and after the fault. The sudden change in current after the fault can directly indicate the power impact near the fault point. The voltage deviation value: monitors the degree to which the voltage of each node deviates from the rated value. Voltage sag is one of the most intuitive electrical characteristics of a fault. By comparing the data before the fault (normal operation) with the data after the fault (abnormal state), the difference features are extracted to eliminate the interference of normal operation fluctuations.

[0076] By performing voltage sag fluctuation analysis on the fault feature vectors before and after the fault, it is found that the voltage sag amplitude is positively correlated with the distance from the fault point. The closer the distance, the more severe the sag. By comparing the voltage deviations of each node before and after the fault, the nodes with voltage sags exceeding the preset threshold are identified, and the fault - affected area is initially delimited. Moreover, different fault types have different impacts on voltage. For example, single - phase grounding faults usually cause the voltage of the non - fault phase to rise, while three - phase short - circuit faults result in a symmetrical voltage sag of the three phases. The fault type can be assisted in judgment through the voltage fluctuation mode, narrowing the scope of the affected area.

[0077] Kirchhoff's Current Law (KCL) describes the conservation relationship of node currents. Combining with Kirchhoff's Voltage Law (KVL), the circuit equations of the power grid can be established. The impedance matrix (or admittance matrix) is a mathematical expression of the electrical characteristics of the power grid. The matrix elements reflect the impedance relationship between nodes and are used to describe the relationship between the post - fault current distribution and voltage drop.

[0078] Constructing an impedance matrix for the preliminarily determined fault - affected area instead of modeling the entire network can achieve refined modeling of the fault area, reduce the amount of calculation, and improve the analysis efficiency. Among them, the matrix parameters are generated based on the topological structure and line parameters to ensure that the model accurately reflects the actual electrical connection.

[0079] By solving the impedance matrix and combining the current / voltage data collected after the fault, the location and impedance value of the fault point are inversely deduced using a fault analysis algorithm. Specifically, the least - squares method or Newton's iterative method can be used to solve the matrix equation to find the fault point with the highest matching degree with the measured data.

[0080] According to the calculation results of the impedance matrix, the specific ranges of the current - over - limit lines and voltage - over - limit nodes caused by the fault are determined.

[0081] In one embodiment, a Phasor Measurement Unit (PMU) is installed at the generator outlet. Using the GPS synchronous clock, nanosecond - level synchronization of the whole - network data is realized, and transient high - frequency data such as power angle, voltage phasor, and frequency are directly collected. The Kalman filtering algorithm is used to denoise the original PMU data and extract the characteristic quantities of transient fluctuations, so as to obtain the transient fluctuation data of the generator set.

[0082] In another embodiment, intelligent meters / Load Monitoring Devices (LMDs) are installed on the low - voltage side of distribution transformers and important load distribution boxes to collect load data such as three - phase current, voltage, and power factor. The fuzzy clustering algorithm is used to classify the loads, and dynamic response models are established respectively. Through state estimation technology, weighted processing is performed on several response models to obtain the load dynamic response quantity.

[0083] In another embodiment, an intelligent terminal (RTU / FTU) is configured for the reactive power compensation device, and the breaker position signal, the switching times of the capacitor bank, and the operation status data of the static var generator (SVG) are collected through hard wiring.

[0084] The original data of the distribution communication network often has quality defects due to problems such as sensor accuracy, communication interference, and device heterogeneity. The preprocessing is performed through a combination of operations including cleaning and denoising, integration and alignment, transformation and efficiency improvement, and specification and dimensionality reduction. Specifically, after determining the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation device status data of the target distribution communication network when a fault occurs, the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation device status data are preprocessed. The steps of the preprocessing include data cleaning, data integration, data transformation, and data specification.

[0085] Specifically, it includes using the Global Navigation Satellite System (GNSS) to time all the acquisition devices to ensure that the timestamp error of the generator set, load, and reactive power equipment data is ≤1ms; alternatively, the three types of data can be mapped into a unified Geographic Information System (GIS) through the distribution network digital twin model to intuitively display the status distribution of each device in the fault-affected area; the triple modular redundancy technology (TMR) is used to collect key data (such as generator power angle, node voltage) from multiple sources, and then outliers are removed through cross-comparison.

[0086] In step S2, dynamic expansion estimation processing is performed on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation device status data to obtain fault impact domain data. Specifically, it includes:

[0087] Based on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation device status data, the fault impact range is traversed to obtain the fault propagation path; a deep learning algorithm is used to extract the fault propagation path to obtain the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude; dynamic expansion estimation processing is performed on the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude to obtain the fault impact domain data.

[0088] Specifically, using the transient fluctuation data of the generator set (such as power angle, speed change), the load dynamic response quantity (such as sudden change in active / reactive power), and the reactive power compensation device status (such as capacitor switching status), the initial fault position and the energy flow direction are comprehensively judged from three key links of the power source, load, and compensation device. By traversing the determined fault impact range, the conduction path of the power fluctuation, voltage / frequency offset caused by the fault in the distribution network is traced.

[0089] Furthermore, a deep learning algorithm is used to extract the fault propagation path, mainly extracting the voltage amplitude change rate, frequency deviation change rate, and power fluctuation amplitude. Specifically, the voltage amplitude change rate reflects the speed of voltage drop or sudden rise of the node voltage after a fault and is used to judge the equipment tolerance risk; the frequency deviation change rate reflects the degree of disruption of the active power balance in the system. A rapid frequency change will trigger the action of the generator governor or the start of the under-frequency load shedding device, while the power fluctuation amplitude can quantify the amplitude of the sudden change in active / reactive power caused by the fault and directly affect the transient stability of the system.

[0090] Based on the change rates and amplitudes of voltage, frequency, and power, combined with the distribution network topology structure, the spatial boundary affected by the fault is calculated through dynamic state estimation. Preferably, the method of dynamic expansion estimation is a real-time correction method based on state estimation.

[0091] In step S2, first, the actual path of fault propagation is determined by traversing multi-source data, and a spatial mapping of "data - line" is established. Then, the physical phenomenon is transformed into quantifiable dynamic features, such as change rates and fluctuation amplitudes, through an algorithm. Finally, the characteristic quantities are transformed into specific influence ranges and degrees through dynamic estimation.

[0092] In step S3, a provincial prediction cascade model is constructed with the fault impact domain data. Among them, the provincial prediction cascade model is configured to process the provincial prediction cascade model according to the matrix factorization method and obtain the cascade reaction prediction result output by the provincial prediction cascade model based on the deep time series network algorithm. Specifically, it includes:

[0093] Construct a fault impact domain feature matrix based on the fault impact domain data; decompose and process the fault impact domain feature matrix using the non-negative matrix factorization method to obtain the associated feature impact vector; use the deep time series network algorithm to model the cascade reaction of the associated feature impact vector to obtain the cascade reaction prediction result.

[0094] Specifically, the fault impact domain data, such as node voltage change rate, power fluctuation amplitude, and equipment status, are converted into a mathematical matrix to form an input format that can be processed by the model.

[0095] The rows and columns of the matrix represent different equipment nodes and characteristic indicators respectively. For example:

[0096] Rows: represent equipment such as buses, lines, transformers, etc.;

[0097] Columns: represent characteristics such as voltage amplitude, frequency deviation, active / reactive power fluctuation, etc.

[0098] Meanwhile, the matrix elements can contain data from different time sections, realizing the multi-dimensional information fusion of space (equipment nodes) and time (fault development process). Moreover, through the numerical magnitudes of the matrix elements, the correlation strength between different devices and features is quantified, such as the voltage correlation between the fault point and adjacent nodes, providing a quantitative basis for the cascading reaction analysis.

[0099] Non-negative matrix factorization (NMF) decomposes a high-dimensional feature matrix into the product of two low-dimensional non-negative matrices. Through the decomposition, the essential features of the data are extracted. At the same time, the non-negativity constraint of non-negative matrix factorization (NMF) makes its results physically interpretable. For example, the column vectors can represent "fault propagation paths", and the row vectors can represent "feature co-variation patterns".

[0100] The deep time series network algorithm can be used to process the correlated sequence data of column vectors and row vectors, capture the dynamic evolution law of the fault cascading reaction, model the dynamic evolution law of the cascading reaction, and obtain the cascading reaction prediction result.

[0101] In S4, the cascading reaction prediction result is subjected to adaptive processing of protection settings based on the impedance trajectory to obtain provincial protection instructions, specifically including:

[0102] Using wavelet transform to extract the cascading reaction prediction result to obtain a voltage fluctuation sequence; performing dynamic mode decomposition processing on the voltage fluctuation sequence to obtain a dominant mode matrix; performing mapping processing on the dominant mode matrix to obtain a power grid node admittance matrix; using the adaptive algorithm for protection settings based on the impedance trajectory to process the power grid node admittance matrix to obtain the provincial protection instructions.

[0103] Specifically, using wavelet transform to extract the cascading reaction prediction result to obtain a voltage fluctuation sequence. Among them, the cascading reaction prediction result contains multi-dimensional information of the power grid operation state, such as voltage, current, power, etc. Wavelet transform is a mathematical tool suitable for the analysis of non-stationary signals and can perform multi-scale decomposition of signals in the time-frequency domain. Therefore, by extracting the characteristic components of voltage fluctuations from the complex prediction result, the voltage change patterns at different frequencies (such as high-frequency transient fluctuations, low-frequency oscillations, etc.) are separated to form a time-series voltage fluctuation sequence. Voltage fluctuation is a direct reflection of power grid faults or abnormal operations. Through wavelet transform, noise interference can be removed and the true fluctuation characteristics can be retained.

[0104] Performing dynamic mode decomposition (DMD) processing on the voltage fluctuation sequence to obtain a dominant mode matrix. In this step, the voltage fluctuation sequence is decomposed into a series of mode components, each mode corresponding to a specific frequency, damping ratio, and spatial distribution. By screening the modes with larger amplitudes, the dominant mode matrix is extracted, where the dominant mode matrix is the oscillation mode that has the greatest impact on the system stability.

[0105] Then, perform mapping processing on the dominant modal matrix to obtain the admittance matrix of the power grid nodes. The admittance matrix is the core parameter matrix that describes the relationship between the node voltages and the injected currents in the power grid. Its elements reflect the electromagnetic coupling characteristics between nodes. In this step, through a mathematical mapping model, such as a system identification method based on modal parameters, convert the characteristics such as frequency and damping of the dominant mode into changes in the admittance values between the power grid nodes. Specifically, the modal parameters can reflect the dynamic characteristics of the system impedance / admittance. Update the admittance matrix through a mapping algorithm to make it represent the power grid structure under the current oscillation mode.

[0106] Use the adaptive protection setting algorithm based on the impedance trajectory to process the admittance matrix of the power grid nodes to obtain provincial protection instructions. Specifically, convert the admittance matrix into an impedance matrix (the inverse matrix of the admittance matrix) to obtain the impedance parameters between each node; based on the impedance trajectory algorithm, analyze the variation law of the impedance trajectory under different fault scenarios, and combine the current operation mode of the power grid (such as load distribution, generator output) to dynamically calculate the setting values such as the action threshold and time limit of the protection device; integrate the calculation results of each region of the provincial power grid to generate unified protection adjustment instructions, such as modifying the setting impedance of the distance protection and the action current of the overcurrent protection. Preferably, the provincial protection instructions include: fault equipment identification, power grid operation mode, post-fault power grid risk, disposal measures, and special requirements during fault disposal.

[0107] In step S4, extract the voltage fluctuation characteristics through wavelet transform, use DMD to identify the dominant oscillation mode affecting system stability, convert the modal characteristics into the real-time admittance matrix of the power grid to reflect the dynamic operation state, and then generate adaptive protection instructions based on the real-time admittance matrix and the impedance trajectory algorithm to achieve a closed-loop control of "state perception - feature extraction - parameter update - protection adjustment".

[0108] In step S5, use the Monte Carlo simulation algorithm to process the obtained topological data of the prefecture-level distribution network and the real-time operation data of the prefecture-level to obtain local system response action instructions, specifically including:

[0109] Use the Monte Carlo simulation algorithm to process the obtained topological data of the prefecture-level distribution network and the real-time operation data of the prefecture-level to obtain a fault scenario database; according to the fault scenario database, perform forward recursive processing on the transient fluctuation data of the generator sets, the dynamic response quantity of the load, and the state of the reactive power compensation equipment to obtain the local system response action instructions.

[0110] Specifically, the Monte Carlo simulation quantifies and analyzes the uncertainty of complex systems through random sampling and probability statistics methods. In the prefecture-level distribution network, the occurrence location, type, and influence range of faults are random, and it is necessary to evaluate the system vulnerability by simulating a large number of possible fault scenarios.

[0111] Based on the distribution network topology (such as line connection relationships, equipment parameters) and real-time operating status (such as load distribution, power flow direction), randomly simulate different types of faults (such as short circuits, line breaks, equipment outages, etc.); through tens of thousands of simulations, covering the occurrence probabilities, location distributions, and time randomness of various types of faults, generate a fault scenario database containing parameters such as fault location, type, occurrence time, and duration.

[0112] Apply the fault scenarios to the power system model, and through the generator rotor motion equation and electromagnetic transient equation, calculate the dynamic changes of parameters such as generator speed, power angle, and output power after the fault, and then conduct transient fluctuation analysis of the generator set; considering the voltage / frequency characteristics of the load, such as the mechanical inertia of motor loads and the temperature characteristics of air-conditioning loads, calculate the dynamic changes of load power after the fault through the load model, where the load model is determined according to the specific grid load; then simulate the dynamic responses of devices such as static var compensators and static var generators after the fault, and calculate their impacts on the system voltage stability; based on the fault scenarios, through forward recursive simulation of the dynamic interactions of the generator set, load, and reactive power devices, and according to the system response results, generate executable local protection instructions.

[0113] Compared with the traditional "post-event processing" mode, the method combining Monte Carlo simulation and forward recursion can more comprehensively evaluate system risks.

[0114] In response to the provincial protection instruction, correct the local system response action instruction, and control the target distribution communication network to execute a strategy matching the correction result.

[0115] The provincial protection instruction represents the global control strategy at the provincial power grid level. Its core goal is to coordinate the actions of protection devices across the province and avoid cross-regional chain reactions caused by local faults. The action instructions of the local system (prefectural-level distribution network) need to be compatible with the provincial strategy to ensure the consistency of the whole network protection logic. However, during the correction process, the parameters, timing, or execution logic of the local instruction need to be adjusted in combination with the goal of the provincial instruction.

[0116] After controlling the target distribution communication network to execute a strategy matching the correction result, the provincial and local coordinated distribution communication network control method further includes: verifying the strategy using the cross-validation method to evaluate the accuracy of the strategy.

[0117] In the provincial and local coordinated control process, the formulation of the strategy depends on the power grid model, fault simulation, and coordination of superior and subordinate instructions. However, in actual execution, there may be model errors, communication delays, or equipment response deviations. The cross-validation method evaluates the effectiveness and robustness of the strategy through multi-dimensional data comparison and scenario reproduction to ensure its reliable execution in the real power grid environment.

[0118] Specifically, the historical operation data and the fault scenarios generated by Monte Carlo simulation can be divided into a training set and a test set. Among them, the training set is used for policy formulation, and the test set is used for independent verification. Select extreme fault scenarios that are not involved in policy formulation to test the generalization ability of the policy.

[0119] If there is a deviation between the verification result and the expectation, the model parameters can be adjusted, the communication protocol can be optimized, or the control logic can be corrected for the error source to correct the policy, and the corrected policy can be re-input into cross-verification until the error converges to an acceptable range.

[0120] Another embodiment of the present invention provides a provincial and local collaborative distribution communication network control system. Specifically, please refer to Figure 2 , Figure 2 which is shown as the structural schematic diagram of the provincial and local collaborative distribution communication network control system in one embodiment of the present invention. The system includes:

[0121] An acquisition module 11, configured to determine the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data of the target distribution communication network when a fault occurs;

[0122] An expansion module 12, configured to perform dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response amount, and the reactive power compensation equipment status data to obtain fault influence domain data;

[0123] A construction module 13, configured to construct a provincial prediction cascade model with the fault influence domain data, where the provincial prediction cascade model is configured to process the provincial prediction cascade model according to the matrix decomposition method and obtain a cascade reaction prediction result output by the provincial prediction cascade model based on the deep time series network algorithm;

[0124] An optimization module 14, configured to perform adaptive processing of protection setting values based on impedance trajectories on the cascade reaction prediction result to obtain provincial protection instructions;

[0125] A simulation module 15, configured to process the obtained prefecture-level distribution network topology data and prefecture-level real-time operation data by using the Monte Carlo simulation algorithm to obtain local system response action instructions;

[0126] An execution module 16, configured to correct the local system response action instructions in response to the provincial protection instructions and control the target distribution communication network to execute a policy matching the correction result.

[0127] Compared with the prior art, the beneficial effects of the embodiments of the present invention are at least one of the following:

[0128] The present invention determines the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment state data of the target distribution communication network when a fault occurs; performs dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment state data to obtain fault influence domain data; constructs a provincial prediction chain model with the fault influence domain data, wherein the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix factorization method, and obtain a cascade reaction prediction result output by the provincial prediction chain model based on the deep time series network algorithm; performs adaptive processing of protection setting values based on the impedance trajectory on the cascade reaction prediction result to obtain a provincial protection instruction; uses the Monte Carlo simulation algorithm to process the obtained topological data and real-time operation data of the prefecture-level distribution network to obtain a local system response action instruction; in response to the provincial protection instruction, corrects the local system response action instruction, and controls the target distribution communication network to execute a strategy matching the correction result.

[0129] Compared with the prior art, the present invention captures the fluctuation data and states when a fault occurs, quantifies the fault influence range, and provides input data in the spatial dimension for subsequent cascade reaction prediction at the provincial level; then reduces the dimensionality of high-dimensional power grid data through the matrix factorization method, extracts key features, and identifies key nodes in the provincial power grid that are prone to trigger cascade faults. Then, through the deep time series network algorithm, the time series features of the data after the fault are mined to predict the possible cascade reactions caused by the fault, anticipate the spread path and potential consequences of the fault in advance at the provincial level, and perform coordinated adjustment of the protection devices within the scope of the provincial power grid according to the predicted cascade reaction risk to obtain a provincial protection instruction; the Monte Carlo simulation can generate a large number of possible operation scenarios through random sampling, generate targeted local system response action instructions based on the specific topology and real-time state of the prefecture-level power grid, correct the local system response action instructions in response to the provincial protection instruction, and control the target distribution communication network to execute a strategy matching the correction result, generate a flexible response strategy based on the local instruction corrected under the provincial instruction, and ensure the compatibility of local control and global stability; through this processing flow, the present invention can improve the fault self-healing ability of the distribution communication network through provincial and local coordination and ensure the safe operation of the power system.

[0130] The above embodiments only represent several implementation manners of the present invention, and the description thereof is relatively specific and detailed, but it should not be construed as a limitation on the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the appended claims.

Claims

1. A control method for a distribution communication network with land-saving and collaborative features, characterized in that, Including: Determine the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment status data of the target distribution communication network when a fault occurs; Perform dynamic expansion estimation processing on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment status data to obtain fault influence domain data; Construct a provincial prediction chain model with the fault influence domain data, where the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix decomposition method and obtain the chain reaction prediction result output by the provincial prediction chain model based on the deep time series network algorithm; Perform adaptive processing of protection setting values based on the impedance trajectory on the chain reaction prediction result to obtain a provincial protection instruction, including: Use wavelet transform to extract the chain reaction prediction result to obtain a voltage fluctuation sequence; Perform dynamic mode decomposition processing on the voltage fluctuation sequence to obtain a dominant mode matrix; Perform mapping processing on the dominant mode matrix to obtain a power grid node admittance matrix; Use the protection setting value adaptive algorithm based on the impedance trajectory to process the power grid node admittance matrix to obtain the provincial protection instruction; Adopt the Monte Carlo simulation algorithm to process the obtained topology data and real-time operation data of the prefecture-level distribution network to obtain a local system response action instruction; In response to the provincial protection instruction, correct the local system response action instruction, and control the target distribution communication network to execute a strategy matching the correction result.

2. The method for controlling a provincial and local collaborative power distribution communication network according to claim 1, wherein Before determining the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment status data of the target distribution communication network when a fault occurs, the provincial and local coordinated control method for the distribution communication network further includes: Determine the fault influence range; The determination of the fault influence range includes: Obtain the topology structure of the target distribution communication network; Collect the load current values and voltage deviation values of several nodes in the topology structure of the target distribution communication network before and after the fault occurs; Extract features from all the load current values and all the voltage deviations before the fault occurs to obtain a first fault feature vector; Extract features from all the load current values and all the voltage deviations after the fault occurs to obtain a second fault feature vector; Perform voltage sag fluctuation analysis on the first fault feature vector and the second fault feature vector to obtain the fault influence area; Construct an impedance matrix of the fault influence area using Kirchhoff's current law, and solve the impedance matrix of the fault influence area to obtain the fault influence range.

3. The control method of the provincial and local collaborative distribution communication network according to claim 2, characterized in that The dynamic expansion estimation processing of the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment status to obtain fault influence domain data includes: Based on the transient fluctuation data of the generator set, the load dynamic response quantity, and the reactive power compensation equipment status data, traverse the fault influence range to obtain a fault propagation path; Use a deep learning algorithm to extract the fault propagation path to obtain the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude; Perform dynamic expansion estimation processing on the voltage amplitude change rate, the frequency deviation change rate, and the power fluctuation amplitude to obtain fault impact domain data.

4. The method for controlling a provincial and local collaborative distribution communication network according to claim 1, characterized in that Construct a provincial prediction chain model with the fault impact domain data, where the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix decomposition method and obtain the prediction result of the chain reaction output by the provincial prediction chain model based on the deep time series network algorithm, including: Construct a fault impact domain feature matrix based on the fault impact domain data; Use the non-negative matrix decomposition method to decompose the fault impact domain feature matrix to obtain the associated feature impact vector; Use the deep time series network algorithm to perform chain reaction modeling on the associated feature impact vector to obtain the prediction result of the chain reaction.

5. The method for controlling a provincial and local collaborative distribution communication network according to claim 1, wherein The Monte Carlo simulation algorithm is used to process the obtained topological data and real-time operation data of the prefecture-level distribution network to obtain local system response action instructions, including: Use the Monte Carlo simulation algorithm to process the obtained topological data and real-time operation data of the prefecture-level distribution network to obtain a fault scenario database; According to the fault scenario database, perform forward recursion processing on the transient fluctuation data of the generator set, the dynamic response of the load, and the state of the reactive power compensation device to obtain the local system response action instructions.

6. The method for controlling a provincial and local collaborative power distribution communication network according to claim 1, wherein The provincial protection instructions include: Fault device identification, power grid operation mode, post-fault power grid risk, disposal measures, and special requirements during fault disposal.

7. The method for controlling a provincial and local collaborative power distribution communication network according to claim 1, wherein, Before performing dynamic expansion estimation processing on the transient fluctuation data of the generator set, the dynamic response of the load, and the state data of the reactive power compensation device, the provincial and local collaborative distribution communication network control method further includes: Perform preprocessing on the transient fluctuation data of the generator set, the dynamic response of the load, and the state data of the reactive power compensation device. The steps of the preprocessing include data cleaning, data integration, data transformation, and data reduction.

8. The method for controlling a provincial and local collaborative power distribution communication network according to claim 1, wherein After controlling the target distribution communication network to execute the strategy matching the correction result, the provincial and local collaborative distribution communication network control method further includes: Use the cross-validation method to verify the strategy to evaluate the accuracy of the strategy.

9. A control system for a distribution communication network with coordinated land and space saving, characterized by Including: An acquisition module for determining the transient fluctuation data of the generator set, the dynamic response of the load, and the state data of the reactive power compensation device of the target distribution communication network when a fault occurs; An expansion module for performing dynamic expansion estimation processing on the transient fluctuation data of the generator set, the dynamic response of the load, and the state data of the reactive power compensation device to obtain fault impact domain data; A construction module for constructing a provincial prediction chain model with the fault impact domain data, where the provincial prediction chain model is configured to process the provincial prediction chain model according to the matrix decomposition method and obtain the prediction result of the chain reaction output by the provincial prediction chain model based on the deep time series network algorithm; An optimization module, configured to perform adaptive processing on the predicted results of the chain reaction based on impedance trajectories to obtain provincial protection instructions, including: extracting the predicted results of the chain reaction by using wavelet transform to obtain a voltage fluctuation sequence; performing dynamic mode decomposition processing on the voltage fluctuation sequence to obtain a dominant mode matrix; performing mapping processing on the dominant mode matrix to obtain a power grid node admittance matrix; and processing the power grid node admittance matrix by using an adaptive protection setting algorithm based on impedance trajectories to obtain the provincial protection instructions; A simulation module, configured to process the obtained topology data and real-time operation data of the prefecture-level distribution network by using the Monte Carlo simulation algorithm to obtain local system response action instructions; An execution module, configured to, in response to the provincial protection instructions, correct the local system response action instructions, and control the target distribution communication network to execute a strategy matching the correction result.

Images