FLS-SVM-based power distribution system multi-mode operation state evaluation method

Abstract

The invention relates to the technical field of power distribution system operation monitoring, and discloses a power distribution system multi-mode operation state evaluation method based on FLS-SVM, and the method comprises the steps: firstly, providing a power distribution system multi-mode operation state definition; secondly, constructing a multi-modal operation state evaluation index system of the power distribution system, performing multi-dimensional evaluation according to respective characteristics of three levels of'region-grid-area ', and then establishing typical scenes of a normal state, a risk state and a fault state of the power distribution system based on a traditional power system operation state description method; and finally, carrying out real-time evaluation on the power distribution system based on FLS-SVM, and identifying a current system-grid-transformer area operation state. Real-time section operation state evaluation of three levels of power distribution system-grid-transformer area is realized, and effective guidance is provided for risk elimination or fault recovery positioning and control.

Description

Technical Field

[0001] The present invention relates to the technical field of power distribution system operation monitoring, and in particular to a multi-modal operation status evaluation method for a power distribution system based on FLS-SVM. Background Art

[0002] The multimodal operating status evaluation index system of the distribution system is the basis for risk prediction and fault recovery control of the power system. Existing evaluation indicators mostly focus on single-level aspects such as power quality, flexibility, and safety, but there is a lack of research on multi-level real-time status evaluation. This application comprehensively considers the characteristics and correlations of the distribution "region-grid-station" at all levels to construct a multi-level, multi-dimensional real-time operating status evaluation index system.

[0003] Data show that in terms of power system operation status classification algorithms, traditional machine learning methods such as support vector machines, decision trees, and K-nearest neighbor algorithms have good generalization capabilities and low computational overhead, and are suitable for processing small-scale and simple-structured problems; while deep learning methods such as CNN, LSTM, and autoencoders perform well in processing complex, nonlinear, and large-scale data, but require higher computing resources. Ensemble learning methods such as random forests and GBDT can provide stronger robustness and accuracy by combining the advantages of multiple models. The fuzzy least squares vector machine classification algorithm combines the advantages of fuzzy logic and least squares support vector machines, and can show strong capabilities in processing data with ambiguity, uncertainty, and complex relationships.

[0004] Chinese patent document CN118690980A discloses a "method and system for evaluating the operating status of a distribution station based on integrated learning". It includes: collecting qualitative and quantitative data of the distribution station; preprocessing the collected qualitative and quantitative data; building a distribution station operating status evaluation model, including: building a distribution station image modal evaluation sub-model, building a distribution station sound modal evaluation sub-model, building a transformer timing evaluation sub-model, and building a modal integration sub-model; obtaining the qualitative and quantitative data of the current operation of the distribution station, and integrating and evaluating the preprocessed time period image features, time period audio features, and transformer timing features to obtain the operating status of the distribution station. The above technical solution is obviously insufficient for multi-level operating status evaluation of large-scale distributed power sources connected to new distribution systems. Summary of the invention

[0005] The present invention mainly solves the technical problem that the existing technical solution is obviously insufficient in multi-level operation status evaluation of large-scale distributed power sources connected to the new distribution system, and provides a distribution system multi-modal operation status evaluation method based on FLS-SVM. First, the definition of the multi-modal operation status of the distribution system is proposed. Secondly, a multi-modal operation status evaluation index system of the distribution system is constructed, and a multi-dimensional evaluation is carried out according to the characteristics of the three levels of "region-grid-substation". The regional layer considers four dimensions of steady-state frequency over-limit, real-time controllability, main transformer load rate and overall voltage level, and proposes five evaluation indicators; the grid layer considers three dimensions of local voltage level, emergency transfer capacity and line load level, and proposes three evaluation indicators; the substation layer considers three dimensions of flexibility level, voltage deviation over-limit and distribution transformer load level, and proposes four three-level evaluation indicators. Then, based on the traditional power system operation status description method, typical scenarios of normal state, risk state and fault state of the distribution system are established. Finally, the distribution system is evaluated in real time based on FLS-SVM to identify the current "region-grid-substation" operation status.

[0006] The above technical problem of the present invention is mainly solved by the following technical solution: The present invention comprises the following steps: S1. Construct a multi-modal operating status evaluation index system for the distribution system, and conduct a multi-dimensional evaluation based on the hierarchical characteristics of "region-grid-station"; S2. Establish typical scenarios of normal state, risk state and fault state of power distribution system; S3. Conduct real-time evaluation of the distribution system, collect operating data at each level of the distribution system as training samples, and use the four operating status labels of the three levels of "region-grid-substation" as output vectors. According to the operating status parameters of the distribution system, a fuzzy rule base is constructed for fuzzy reasoning, distinguishing different operating status categories at each level and identifying the current "region-grid-substation" operating status.

[0007] Preferably, the multimodal operating states of the power distribution system include: the normal state is the normal operating state of the power distribution system, the risk state is the warning state during system operation, the fault state is the operating state of the system when a fault occurs, and the recovery state is the process of gradually restoring the power supply of the distribution system to normal after the fault state occurs.

[0008] Preferably, the regional layer of "region-grid-station" refers to the entire regional scope including the main substation distribution system, and the evaluation indicators at the distribution area level reflect the impact of system frequency, real-time controllability, overall voltage level and main transformer load level.

[0009] Preferably, the grid layer of "region-grid-substation" refers to a medium-voltage feeder in the distribution system, which is responsible for the power transmission network from the high-voltage power grid to the low-voltage users. The evaluation indicators at the grid level reflect the impact of local voltage levels, emergency transfer capabilities and line load levels.

[0010] Preferably, the substation layer of the "region-grid-substation" refers to the area covered by a distribution transformer, which is responsible for distributing electricity to low-voltage users. The evaluation indicators at the distribution substation level reflect the impact of flexibility level, voltage deviation limit and distribution transformer load level.

[0011] As a preference, multimodal operating status evaluation indicators of some distribution systems are established based on risk preference utility functions. Regional layer evaluation indicators include frequency upper limit risk and frequency lower limit risk, as well as net load fluctuation rate, regional voltage qualification rate and main transformer over-limit risk.

[0012] Preferably, the grid-layer evaluation indicators include local voltage qualification rate, emergency power transfer margin and line over-limit risk; the substation-layer evaluation indicators include voltage over-upper-limit risk and voltage over-lower-limit risk, as well as flexibility margin and distribution transformer over-limit risk.

[0013] Preferably, the normal state, risk state and fault state are identified based on whether all the typical operation scenario rules meet the quantitative classification, and the four distribution system operation states including the recovery state are comprehensively identified by combining fuzzy rules and least squares support vector machine typical scenarios.

[0014] As a preferred method, a fuzzy membership function model FLS-SVM based on LS-SVM is adopted. The FLS-SVM model combines fuzzy rules and LS-SVM at the same time, classifies data according to the distance from the data point to the classification hyperplane, and describes the different operating states of the system through fuzzy rules, avoiding the requirements of traditional classification methods for precise boundaries and adapting to the complexity of multimodal systems.

[0015] Preferably, the real-time evaluation of the power distribution system specifically includes: S3.1. Collecting operation data of each level of the distribution system as training samples x , the data of each sample are the measured parameters of each level of the distribution system at the same time, and the training samples are respectively labeled with four operating states at three levels of "region-grid-station" as output vectors; the measured parameters of each level (such as node voltage, line transmission power, transformer load rate, frequency and other parameters required by the evaluation index system.

[0016] S3.2. Normalize the original training samples; S3.3. According to the operating status parameters of the distribution system, a fuzzy rule base is constructed and the input data is fuzzified; S3.4. After fuzzy reasoning, the safety status evaluation value obtained is used as a feature input into the SVM model; S3.5. Use the support vector machine algorithm to train the training data set and find the optimal hyperplane to distinguish the different operating status categories at each level; S3.6. Input the real-time collected distribution system operation data into the trained FLS-SVM multi-classification model to determine the operating status categories of each level of the distribution network based on the current time section measurement data.

[0017] The beneficial effects of the present invention are: 1. The FLS-SVM multi-classification model is used to realize the real-time section operation status evaluation of the three levels of power distribution: "region-grid-station", providing effective guidance for risk elimination or fault recovery positioning and control; 2. Based on the characteristics of each level of the new power distribution system, a multi-dimensional evaluation index system suitable for each level is proposed to accurately evaluate the real-time operating status of each level and the system as a whole; 3. The FLS-SVM multi-classification model is used as a new distribution state division algorithm to improve the calculation speed and accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a flow chart of the present invention.

[0019] Figure 2 It is a schematic diagram of the multi-level topological relationship of power distribution "region-grid-station area" of the present invention.

[0020] Figure 3 It is a multi-modal operation status analysis diagram of a power distribution system of the present invention.

[0021] Figure 4 It is a multi-modal operating state and control diagram of a power distribution system of the present invention.

[0022] Figure 5 It is a distribution network multi-modal operation status evaluation index system diagram of the present invention.

[0023] Figure 6 This is an example diagram of a "region-grid-station area" network topology calculation according to the present invention. DETAILED DESCRIPTION

[0024] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the technical solutions of the present application are further described in detail below through embodiments and in combination with the accompanying drawings. It should be understood that the specific implementation method described here is only an optimal embodiment of the present application, which is only used to explain the present application and does not limit the protection scope of the present application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0025] This application proposes a new real-time evaluation model for the multimodal operation status of a distribution system based on FLS-SVM. First, a definition of the multimodal operation status of a distribution system is proposed. Secondly, an evaluation index system for the multimodal operation status of a distribution system is constructed, and a multi-dimensional evaluation is performed according to the characteristics of the three levels of "region-grid-substation". The regional level considers four dimensions of steady-state frequency over-limit, real-time controllability, main transformer load rate, and overall voltage level, and proposes five evaluation indicators; the grid level considers three dimensions of local voltage level, emergency transfer capacity, and line load level, and proposes three evaluation indicators; the substation level considers three dimensions of flexibility level, voltage deviation over-limit, and distribution transformer load level, and proposes four three-level evaluation indicators. Then, based on the traditional power system operation status description method, typical scenarios of normal state, risk state, and fault state of the distribution system are established. Finally, the distribution system is evaluated in real time based on FLS-SVM to identify the current "region-grid-substation" operation status.

[0026] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the operations (or steps) as sequential processes, many of the operations (or steps) therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the drawings; the process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0027] The technical solution of the present invention is further specifically described below through embodiments and in conjunction with the accompanying drawings.

[0028] Embodiment: The multi-modal operation status evaluation method of the power distribution system based on FLS-SVM in this embodiment is as follows: Figure 1 As shown. Based on the idea of ​​"horizontal and vertical coordination, hierarchical and regional" control of the power distribution system, this application divides the power distribution system into three levels: "region-grid-station" and analyzes them level by level. The topological diagram of the multi-level relationship of "region-grid-station" in power distribution is shown as follows: Figure 2 shown.

[0029] The regional layer usually refers to the entire area of ​​the main substation distribution system. It relies on EMS, DMS, and multi-state control systems to formulate main and distribution coordinated control strategies to achieve regional optimization. Therefore, the evaluation indicators at the distribution area level mainly reflect the impact of system frequency, real-time controllability, overall voltage level, and main transformer load level.

[0030] The grid layer refers to a medium-voltage feeder in the distribution system, which is mainly responsible for the power transmission network from the high-voltage power grid to the low-voltage users. It relies on the distribution automation master station system, uses adjustable resources, responds to the main grid dispatch, participates in the main distribution coordinated control, and realizes grid mutual assistance. Therefore, the evaluation indicators at the grid level mainly reflect the impact of local voltage levels, emergency transfer capabilities, and line load levels.

[0031] The substation layer refers to the area covered by a distribution transformer, which is responsible for distributing electricity to a certain number of low-voltage users and is the "last mile" of power distribution. Relying on the fusion terminal, the low-voltage substation can achieve local balance of source, network, load and storage, respond to distribution network scheduling, and participate in distribution network interaction and regional autonomy. Therefore, the evaluation indicators at the distribution substation level mainly reflect the impact of flexibility level, voltage deviation exceeding the limit, and distribution transformer load level.

[0032] In summary, the regional layer, grid layer and substation layer influence each other, forming a control system architecture of regional optimization, grid mutual assistance and substation autonomy. When evaluating the multi-modal operation status of the distribution system, it is necessary to realize the operation status division by levels.

[0033] According to the operation and control characteristics of the distribution area-grid-station area, it is necessary to use measurement data and boundary conditions to identify and divide the safe state of the distribution system to provide theoretical support for the reliability and safety analysis of the distribution system. According to the multi-modal operation state of the distribution system, comprehensive evaluation indicators and efficient decision-making methods can be determined to accurately find energy efficiency improvement points and effectively manage the distribution system.

[0034] According to the operating state and inequality constraints of the new distribution system, the multimodal operating state of the distribution system can be divided into normal state, risk state, fault state and recovery state. The four state definitions, optimization measures and optimization objectives are as follows: Figure 3 shown.

[0035] The relationship between the four operating states of the distribution system and each control state is as follows: Figure 4 shown.

[0036] The normal state is the normal operating state of the distribution system, which is manifested in that the system safety requirements are met and various operating indicators are in a stable state, and can effectively resist various foreseeable disturbances in the future.

[0037] The risk state is a warning state in system operation, which is mainly manifested in low system capacity margin, poor overall voltage level, and excessive voltage deviation. When the system's future state load or distributed power source fluctuates greatly, the system in the risk state has poor ability to resist disturbances, and the system is more likely to directly transform from the risk state to the fault state, causing some power quality indicators to exceed the limit. Therefore, when the system is in a risk state, it should be promptly converted to a normal state through multi-resource correction control to enhance the system's ability to resist risks.

[0038] Fault state usually refers to the operating state of the system under fault or abnormal conditions. The occurrence of fault state means that some parts, components or connections of the system have failed, causing the system to fail to operate normally as expected. Fault state has a great impact on the reliability of the power distribution system, and appropriate repair or recovery control measures should be taken in time to ensure the reliability of system load power consumption.

[0039] The recovery state is the process in which the power distribution system gradually restores power supply to normal after the fault state occurs through optimized control, flexible power transfer, etc. The process is specifically reflected in the process of converting from the fault state to the risk state or directly from the fault state to the normal state.

[0040] Construction of a multi-level and multi-dimensional evaluation indicator system In constructing the distribution system multimodal operation status evaluation index system, this application follows the principles of operability, comprehensiveness, and scientific rationality. Operability is reflected in the measurability of indicator variables; comprehensiveness is reflected in the operation status of the three levels of distribution area-grid-substation, reflecting the impact of high-penetration distributed power sources on the three dimensions of distribution system safety, flexibility, and reliability; scientificity is reflected in the construction of quantifiable indicators from different angles according to the characteristics of different levels and dimensions. Distribution system multimodal operation status evaluation index system such as Figure 5 shown.

[0041] Calculation method of multi-modal operation status evaluation index of distribution system The relationship between the operation risk, fault and severity of the problem in the distribution system is often not linear. The more serious the problem is, the more likely the consequences will increase exponentially. To this end, this paper combines the risk preference utility function to establish a multi-modal operation status evaluation index for the distribution system. The risk preference utility function calculation formula is as follows: (1) Where: e is a natural constant; For the i Multi-modal operating status evaluation indicators; Evaluation indicators based on risk preference utility function.

[0042] (1) Regional level evaluation indicators 1) Risk of frequency exceeding upper limit (2) Where: for t Risk of the system frequency exceeding the upper limit at any given moment; is the rated frequency; for t System frequency at the moment; The upper limit of the system frequency.

[0043] 2) Risk of frequency exceeding the lower limit (3) Where: for t The risk of the time frequency exceeding the lower limit; is the lower limit of system frequency; for t System frequency at the moment.

[0044] 3) Net load fluctuation rate (4) Where: For the system t Net load fluctuation rate at each moment; For the Photovoltaic power sources, is the total number of photovoltaic power sources, including centralized photovoltaic power sources and distributed photovoltaic power sources; For the Energy storage system, is the total number of energy storage systems; , They are the ramp power allowed for photovoltaic power grid connection and the ramp power allowed for energy storage; is the climbing power of the system itself; For the system t Net load at the moment.

[0045] 4) Regional voltage qualification rate (5) Where: for t The qualified rate of regional voltage at the moment; is the number of regional stations; for t The number of substations with qualified voltage at any moment.

[0046] 5) Main transformer over-limit risk (6) Where: for t Main change at all timesa Risk of over-limit; Main change a The maximum permissible transmission capacity; for t Main change at all times a When calculating the power flow of the distribution network, the main transformer takes a positive value when transmitting power to the load in the area where it is located, and takes a negative value when the main transformer sends power back to the upper grid, so the absolute value is added here.

[0047] (2) Grid layer evaluation indicators 1) Local voltage qualification rate (7) Where: for t Time Grid i Voltage qualification rate; For Grid i The number of stations; for t Time Grid i The number of voltage qualified areas.

[0048] 2) Emergency transfer margin (8) Where: for t Time Grid i Emergency transfer margin; for t Time adjacent grid j The transmission capacity of the line; , They are t Time Grid i and j of load.

[0049] 3) Line over-limit risk (9) Where: for t Time Grid i Line x The risk of exceeding the limit; for t Time Grid i Line x The transmission power of For Grid i Line x When calculating the power flow of the distribution network, the line transmission power is positive when it flows in the forward direction and negative when it flows in the reverse direction, so the absolute value is added here.

[0050] (3) Evaluation indicators at the substation level 1) Flexibility margin (10) (11) (12) Where: for t Time Zone y flexibility margin; , They are positive and negative values ​​of net load change respectively; for t Time Zone y The upward flexibility adjustment ability index; For Taiwan y Energy storage system collection; for t Time Zone y Energy Storage System e The charge of For Taiwan y Energy Storage System e The minimum charge of is the discharge efficiency of the energy storage system; For Taiwan y Energy Storage System e Rated power; for t Time Zone y Energy Storage System e The discharge power. for t Time Zone y Downward flexibility adjustment capability indicator; Charging efficiency for energy storage systems; For Taiwan y Energy Storage System e The maximum charge of; Charge / Charging efficiency = Power, Charge*Charging efficiency and Charge / Discharging efficiency in Formula (11) and Formula (12) are fixed expressions, for t Time Zone y Net load power; is the discharge efficiency of the energy storage system.

[0051] 2) Risk of voltage exceeding upper limit (13) Where: for t Time Zone y node i The voltage exceeds the upper limit risk; is the upper voltage limit; is the nominal voltage; for t Time Zone y node i voltage.

[0052] 3) Risk of voltage exceeding the lower limit (14) Where: for t Time Zone y node i The voltage lower limit safety margin; is the voltage lower limit; for t Time Zone y node i voltage.

[0053] 4) Risk of over-limit of distribution transformer (15) Where: for t Time Zone y Distribution Transformer w Risk of over-limit; For Taiwan y Distribution Transformer w The maximum permissible transmission capacity; for t Time Zone y Distribution Transformer w transmission capacity.

[0054] Construction of typical scenarios of power distribution system operation status The normal state, risk state and fault state of the distribution system can be identified based on whether all the rules of typical operation scenarios are met quantitatively, while the recovery state is a process quantity with certain coupling and needs to be divided according to relevant algorithms. To this end, fuzzy rules and least squares support vector machine (LS-SVM) are combined to identify the four operating states of the distribution system. Considering the operation scenario of the new distribution system, the operation state of the multi-level new distribution system based on the future operation state is comprehensively identified.

[0055] The reference standards for the status classification of some indicators in this article are as follows: (1) According to the power quality requirements for allowable deviation of power system frequency, the normal value of the grid frequency is 50 Hz. For systems with a grid capacity of less than 3 million kilowatts, the frequency deviation should not exceed ±0.5 Hz.

[0056] (2) According to the load guideline for power transformers, when the load rate of the distribution transformer is below 80%, it indicates that the substation is operating normally; when the load rate of the distribution transformer is between 80% and 100%, it indicates that the substation is overloaded; when the load rate of the distribution transformer is higher than 100%, it indicates that the substation is overloaded.

[0057] (3) According to the power quality supply voltage tolerance regulations: the sum of the absolute values ​​of the positive and negative deviations of the supply voltage of 35kV and above shall not exceed 10% of the rated voltage; the allowable deviation of the supply voltage of 10kV and below is ; The allowable deviation of 220kV single-phase power supply voltage is 7%-10%.

[0058] 4.3.1 Normal State Scenario of New Power Distribution System The normal operation scenarios of the new power distribution system are as follows: (1) Frequency: Normally it does not exceed ±0.3 Hz.

[0059] (2) Net load fluctuation: The net load fluctuation rate is less than 0.8.

[0060] (3) Voltage qualification rate: The voltage qualification rate of regional nodes is required to be greater than 0.95; the local voltage qualification rate of the grid layer is required to be greater than 0.98.

[0061] (4) Emergency power transfer margin: There is a power transfer switch between grid layers, and the main transformer and distribution transformer fault risk indicators are in a normal state. If a fault occurs, the adjacent grid can transfer power at a capacity of 1.2 times the load of the grid.

[0062] (5) Distribution transformer load rate: Load rate is less than 80% (6) Flexibility margin: The normal flexibility margin index of the substation is greater than 1.2.

[0063] (7) Voltage deviation: For distribution networks with voltage levels of 10 kV and below, the allowable voltage deviation is less than ±5%.

[0064] According to the normal operation scenario, the state value ranges of formulas (2)-(15) are calculated as shown in Table 1.

[0065] Table 1 Normal operation index range

[0066] New distribution system risk scenarios The risky operation scenarios of the new distribution system are as follows: (1) Frequency: For systems with a grid capacity of less than 3 million kilowatts, the frequency deviation range is [-0.5, -0.3] Hz or [0.3, 0.5] Hz.

[0067] (2) Net load fluctuation: The net load fluctuation rate is between 0.8 and 1.

[0068] (3) Voltage qualification rate: The voltage qualification rate of regional nodes is between 0.9 and 0.95; the local voltage qualification rate of grid layers is between 0.95 and 0.98.

[0069] (4) Emergency power transfer margin: There is a power transfer switch between grid layers, and when a fault occurs, the adjacent grid can transfer power at a capacity of 0.8 to 1.2 times the load of the grid.

[0070] (5) Transformer load factor: The load factor is less than 80%.

[0071] (6) Flexibility margin: The normal flexibility margin index of the substation is between [1,1.2).

[0072] (7) Voltage deviation: For distribution networks with voltage levels of 10 kV and below, the allowable voltage deviation is between 5% and 7% or -7% and -5%.

[0073] According to the risk state operation scenario, the state value ranges of formulas (2)-(13) are calculated as shown in Table 2.

[0074] Table 2 Risk state operation indicator range

[0075] New distribution system fault scenario The fault operation scenarios of the new power distribution system are as follows: (1) Frequency: For systems with a grid capacity of less than 3 million kilowatts, the frequency deviation is greater than 0.5 Hz or less than -0.5 Hz.

[0076] (2) Net load fluctuation: The net load fluctuation rate is greater than 1.

[0077] (3) Voltage qualification rate: The regional node voltage qualification rate is less than 0.9; the local voltage qualification rate of the grid layer is less than 0.95.

[0078] (4) Emergency power transfer margin: There is a power transfer switch between grid layers, and when a fault occurs, the adjacent grid can transfer power at a capacity less than 0.8 times the load of the grid, or there is no power transfer switch between grid layers.

[0079] (5) Transformer load factor: The load factor is higher than 120%.

[0080] (6) Flexibility margin: The normal flexibility margin index of the substation is in the range [0,1).

[0081] (7) Voltage deviation: For distribution networks with voltage levels of 10 kV and below, the allowable voltage deviation is greater than 7% or less than -7%.

[0082] According to the fault state operation scenario, the state value ranges of formulas (2)-(15) are calculated as shown in Table 3.

[0083] Table 3 Fault state operation index range

[0084] Multi-modal operating state identification based on FLS-SVM LS-SVM is an improved support vector machine based on statistical theory, which is mostly used for data regression and binary classification problems. For multi-classification problems, many scholars in recent years have dealt with them by converting them into multiple binary classification problems. Common strategies include "one-to-one" and "one-to-many" multi-classification methods. However, in practical applications, the classification mechanisms of these two methods may lead to inseparable areas between categories, which will affect the correct classification of samples and reduce the classification accuracy. To solve the above problems, this paper adopts the fuzzy membership function model FLS-SVM based on LS-SVM. The FLS-SVM model combines the advantages of fuzzy rules and LS-SVM, and can show strong ability to classify data based on the distance from the data point to the classification hyperplane. The variables for dividing the multimodal operating state of the distribution system are as follows: (16) Where: for t The state variables of the power distribution system at time t, i.e., the input quantities mentioned below.

[0085] In LS-SVM, for the “one-to-many” i The original discriminant function of the classifier Usually it can be written (17) Where: Reason The input quantities constituted are the status evaluation indicators of each level of the power distribution system; is the Lagrange multiplier; is the classification threshold; It is the output category identifier, i.e. the operating status of the power distribution system; is the kernel function.

[0086] In the "one-to-many" case, the output of each classifier is transformed into "Forward mapping" to the interval [0,1] to establish a one-dimensional fuzzy membership function , as shown in formula (18).

[0087] (18) Where: is the smoothing factor, and its value is greater than 0.

[0088] When classifying samples,x Assigned to the category with the largest membership function for (19) When "fuzzy output" is required, keep the entire group .

[0089] This FLS-SVM classifier flexibly describes the different operating states of the system through fuzzy rules, avoiding the requirement of precise boundaries in traditional classification methods, and thus can adapt to the complexity of multimodal systems.

[0090] Training sample feature vector is the formula (17) Composed of The input matrix is n is the sample size, m is the number of eigenvectors. According to the inequality constraints for determining the multi-modal operating state of the distribution system, the eigenvector of the input sample distribution system is determined, as shown in formula (20). The training sample eigenvector is written as a matrix , Sample by submatrix region layer , Grid layer sample , station layer samples constituted.

[0091] (20) Output vector Identify the type of sample: (twenty one) Where: For Region q Sample in t The operating status at the moment; For Grid w Sample in t The operating status at the moment; For Taiwan y Sample in t The operating status at the moment; it is divided into 4 categories according to the operating status of the new distribution system. When the class label is 1, it indicates a normal state; when it is 2, it indicates a risk state; when it is 3, it indicates a fault state; and when it is 4, it indicates a recovery state.

[0092] Preprocessing of sample data In order to avoid the adverse effects of multi-dimensional and high-dimensional sample data in the process of model building, this paper adopts the normalization method to preprocess the training sample data to speed up the sample training speed and convergence speed of the model and improve the prediction accuracy.

[0093] (twenty two) Where: , are the maximum and minimum values ​​of the data samples, respectively.

[0094] The normalization method is used to preprocess the training samples. Sample normalization usually refers to scaling the data to the same range according to the features, such as 0 to 1. After normalization, all features are in a similar numerical range, making the optimization process more stable and accelerating the convergence of the model. At the same time, the fuzzy membership function is not simply binarized here.

[0095] Distribution system safety status assessment process based on FLS-SVM The multi-modal real-time status assessment model of the distribution system consists of two parts: offline training and online application. The specific steps are as follows, and the assessment flow chart is as follows: Figure 1 shown.

[0096] S1: Collecting operation data of each level of the power distribution system as training samples x , the data of each sample are the measured parameters of each level of the distribution system at the same time, and the training samples are labeled with four operating states at three levels of “region-grid-station” as output vectors.

[0097] S2: Normalize the original training samples.

[0098] S3: According to the operating status parameters of the distribution system, a fuzzy rule base is constructed and the input data is fuzzified.

[0099] S4: After fuzzy inference, the safety status assessment value obtained is usually input into the SVM model as a feature.

[0100] S5: Use the support vector machine algorithm to train the training data set. The SVM algorithm distinguishes different operating status categories at each level by finding the optimal hyperplane.

[0101] S6: Input the real-time collected distribution system operation data into the trained FLS-SVM multi-classification model to determine the operation status categories of each level of the distribution network of the current time section measurement data.

[0102] The best implemented power distribution system topology is Figure 6 As shown. The distribution system is a region, which contains 3 grids, and the 3 grids have a total of 12 substations. Grid A has 4 substations, 2 centralized photovoltaic power stations, and 1, and there is a tie switch between it and grid B; grid B has 4 substations and 3 centralized photovoltaic power stations; grid C has 4 substations and 4 centralized photovoltaic power stations.

[0103] The specific embodiments described in this application are merely examples of the spirit of the present invention. The above embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of protection of the present application. It should be pointed out that a technician in the technical field to which the present application belongs may make various modifications or supplements to the described specific embodiments or replace them in a similar manner, but will not deviate from the spirit of the present application or exceed the scope defined in the attached claims. For a person of ordinary skill in the art, multiple variations and improvements may also be made without departing from the concept of the present application. Therefore, the scope of protection of the present application shall be subject to the attached claims.

Claims

1. A multi-modal operating status evaluation method for a power distribution system based on FLS-SVM, characterized in that: The following steps are involved: S1. Construct a multi-modal operating status evaluation index system for the distribution system, and conduct multi-dimensional evaluation based on the hierarchical characteristics of "region-grid-station"; S2. Establish typical scenarios of normal state, risk state and fault state of power distribution system; S3. Conduct real-time evaluation of the distribution system, collect operating data at each level of the distribution system as training samples, and use the four operating status labels of the three levels of "system-grid-substation" as output vectors. According to the operating status parameters of the distribution system, a fuzzy rule base is constructed for fuzzy reasoning, different operating status categories at each level are distinguished, and the current "system-grid-substation" operating status is identified.

2. The multi-modal operating status evaluation method of a power distribution system based on FLS-SVM according to claim 1 is characterized in that: The multimodal operating states of the distribution system include: the normal state is the normal operating state of the distribution system, the risk state is the warning state during system operation, the fault state is the operating state of the system when a fault occurs, and the recovery state is the process of the distribution system gradually restoring power supply to normal after the fault state occurs.

3. The multi-modal operating status evaluation method of a power distribution system based on FLS-SVM according to claim 1 is characterized in that: The regional layer of "region-grid-station" refers to the entire regional scope of the main substation distribution system. The evaluation indicators at the distribution area level reflect the impact of system frequency, real-time controllability, overall voltage level and main transformer load level.

4. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 1 is characterized in that: The grid layer of "region-grid-substation" refers to a medium-voltage feeder in the distribution system, which is responsible for the power transmission network from the high-voltage power grid to the low-voltage users. The evaluation indicators at the grid level reflect the impact of local voltage levels, emergency transfer capabilities and line load levels.

5. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 4 is characterized in that: The substation layer of "region-grid-substation" refers to the area covered by a distribution transformer, which is responsible for distributing electricity to low-voltage users. The evaluation indicators at the distribution substation level reflect the impact of flexibility level, voltage deviation limit and distribution transformer load level.

6. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 5 is characterized in that: Based on the risk preference utility function, the multimodal operating status evaluation indicators of some distribution systems are established. The regional layer evaluation indicators include the risk of exceeding the upper limit of frequency and the risk of exceeding the lower limit of frequency, as well as the net load fluctuation rate, regional voltage qualification rate and main transformer over-limit risk.

7. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 6 is characterized in that: Grid-level evaluation indicators include local voltage qualification rate, emergency power transfer margin and line over-limit risk; substation-level evaluation indicators include voltage upper-limit risk and voltage lower-limit risk, as well as flexibility margin and distribution transformer over-limit risk.

8. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 1 or 2, characterized in that: The normal state, risk state and fault state are identified according to whether all the rules of typical operation scenarios meet the quantitative classification. The four distribution system operation states including the recovery state are comprehensively identified by combining fuzzy rules and least squares support vector machine typical scenarios.

9. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 1 or 2, characterized in that: A fuzzy membership function model FLS-SVM based on LS-SVM is adopted. The FLS-SVM model combines fuzzy rules and LS-SVM at the same time. It classifies data according to the distance from the data point to the classification hyperplane, and describes the different operating states of the system through fuzzy rules, avoiding the requirements of traditional classification methods for precise boundaries and adapting to the complexity of multimodal systems.

10. The method for evaluating the multi-modal operation status of a power distribution system based on FLS-SVM according to claim 1 or 2, characterized in that: Real-time evaluation of the power distribution system includes: S3.

1. Collecting operation data of each level of the distribution system as training samples x , the data of each sample are the measured parameters of each level of the distribution system at the same time, and the training samples are respectively labeled with four operating states at three levels of "system-grid-station" as output vectors; S3.

2. Normalize the original training samples; S3.

3. According to the operating status parameters of the distribution system, a fuzzy rule base is constructed and the input data is fuzzified; S3.

4. After fuzzy reasoning, the safety status evaluation value obtained is input into the SVM model as a feature; S3.

5. Use the support vector machine algorithm to train the training data set and find the optimal hyperplane to distinguish the different operating status categories at each level; S3.

6. Input the real-time collected distribution system operation data into the trained FLS-SVM multi-classification model to determine the operating status categories of each level of the distribution network based on the current time section measurement data.

Images