Broadband oscillation in-situ monitoring and cooperative defense system

Abstract

The invention provides a broadband oscillation in-situ monitoring and cooperative defense system, and relates to the technical field of power system control, and the system comprises a new energy substation which is used for obtaining the electrical parameters of a new energy station, calculating the oscillation characteristic quantity of the new energy station according to the electrical parameters, and achieving the evaluation, early warning and control removal functions of broadband oscillation; the regional master station is used for receiving the electrical parameters and the oscillation characteristic quantity sent by the new energy substation and positioning an oscillation source by using a broadband oscillation monitoring traceability technology so as to realize accurate control in turns; and the dispatching master station is used for monitoring the frequency and voltage level of the power system and the generator tripping quantity information of each regional master station in real time, carrying out system safety check by combining the frequency and voltage level of the system, the generator tripping quantity of each master station and the operation condition of the system, and sending a locking instruction to the related regional master station, thereby realizing effective prevention and control of broadband oscillation on the basis of ensuring the safety of a large power grid. According to the invention, accurate positioning and cooperative prevention and control of broadband oscillation of the novel power system are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of power system control. Specifically, it particularly relates to a wide - frequency oscillation in - situ monitoring and collaborative defense system. Background Art

[0002] With the large - scale integration of new energy into the grid, the formation of high - voltage direct - current transmission networks, and the commissioning of power - electronic loads, the interaction between power - electronic devices and the power grid in the power system can cause wide - frequency oscillations in the frequency range from a few Hz to several thousand Hz. Wide - frequency oscillations may damage power equipment, lead to the outage of new - energy generating units, etc., seriously affecting equipment safety and threatening the stable operation of the system, and becoming an important factor restricting the efficient consumption of new energy. In recent years, wide - frequency oscillation problems have occurred frequently in regions with a high proportion of new energy at home and abroad. Oscillation events in multiple frequency ranges have occurred during the engineering commissioning and operation in some areas with intensive new - energy access.

[0003] Affected by multiple aspects such as the multi - time - scale, strong non - linear characteristics of power - electronic devices, and the complex coupling effects between devices, multi - machine groups in new - energy power stations interact with each other. The generation mechanism, oscillation characteristics, and influencing factors of the propagation range of wide - frequency oscillations are intricate and complex, and have significant characteristics such as a wide - frequency domain, strong time - variability, strong non - linearity, multi - modality, and wide - area propagation. Currently, there is still a lack of a unified and effective mathematical model and analysis method, making it difficult to accurately analyze through off - line simulation and implement effective prevention and control measures. There is an urgent need to carry out on - line monitoring and prevention and control.

[0004] Traditional wide - frequency oscillation monitoring and prevention and control measures focus on on - line identification of oscillation frequencies and amplitudes, but there are deficiencies in accurate tracing. Since the root cause of oscillation problems is difficult to identify, once an oscillation occurs, the conventional approach is to "coarsely" cut off the corresponding power stations after the oscillation amplitude and frequency reach the threshold. With the large - scale centralized access of "desert, Gobi, and wasteland" new energy, the scale of new - energy aggregation access in local areas with strong electrical coupling exceeds tens of millions of kilowatts, and the number of units affected by wide - frequency oscillations increases significantly. Adopting the traditional method of in - situ monitoring and cutting off the oscillating power stations will lead to a large - area disconnection of new energy from the grid, seriously affecting the stable operation of the power grid.

[0005] Regarding the problems in the related technology, no effective solution has been proposed yet. Summary of the Invention

[0006] In view of this, the present invention provides a wide - frequency oscillation in - situ monitoring and collaborative defense system to solve the above - mentioned problems.

[0007] To solve the above problems, the specific technical solutions adopted by the present invention are as follows:

[0008] A wide - frequency oscillation in - situ monitoring and collaborative defense system includes:

[0009] A new energy sub-station is used to obtain the electrical parameters of a new energy power station, calculate the oscillation characteristic quantities of the new energy power station according to the electrical parameters; when the oscillation characteristic quantities reach a preset oscillation threshold, send the electrical parameters and oscillation characteristic quantities to the regional master station, receive and execute the generator tripping instruction issued by the regional master station;

[0010] A regional master station is used to receive the electrical parameters and oscillation characteristic quantities sent by the new energy sub-station, and use broadband oscillation monitoring and tracing technology to locate the oscillation source; select the generators to be tripped according to the oscillation source location result, and send the generator tripping instruction to the new energy sub-station;

[0011] A dispatching master station is used to monitor the power system frequency, voltage level and the generator tripping amount information of each regional master station in real time, perform security checks in combination with the power system frequency, voltage level and the generator tripping amount information of each regional master station, and perform power coordination defense according to the check results.

[0012] Preferably, the new energy sub-station includes:

[0013] A data acquisition and analysis unit is used to obtain the electrical parameters of the new energy power station, and identify the oscillation events and oscillation event patterns in the new energy power station by analyzing the development trend of the electrical parameters;

[0014] A regional master station sending unit is used to send the electrical parameters of the new energy power station to the regional master station after an oscillation occurs in the new energy power station;

[0015] A receiving and executing unit is used to receive and execute the generator tripping instruction issued by the regional master station.

[0016] Preferably, the data acquisition and analysis unit includes:

[0017] A data monitoring unit is used to use an oscillation on-line monitoring and analysis device to monitor and analyze the electrical parameters of a wind farm, a photovoltaic power station generating unit and a feeder in real time based on time-division multiplexing method and adaptive sampling frequency technology, and the electrical parameters include voltage, current, active power, reactive power and apparent power;

[0018] A window characteristic factor extraction unit is used to analyze the voltage-current angle and the power curve fluctuation characteristics according to the electrical parameters of the wind farm, the photovoltaic power station generating unit and the feeder, and extract the window characteristic factors by using data statistics and deduction methods;

[0019] An oscillation event identification unit is used to identify oscillation events based on the window characteristic factors through data statistics and trend analysis, and by integrating the multi-factor weight ratio fusion luminescence method;

[0020] An oscillation event pattern recognition unit is used to perform data augmentation processing on electrical parameters through a data augmentation-based multimodal fusion strategy, and fuse transfer learning and generative adversarial networks to construct an oscillation pattern recognition model, and use the oscillation pattern recognition model to identify oscillation event patterns.

[0021] Preferably, based on the electrical parameters of the wind farm, photovoltaic power station power generation units and feeders, analyze the voltage-current phase angle and power curve fluctuation characteristics, and use data statistics and deduction methods to extract window feature factors to form oscillation risk assessment elements, including:

[0022] Based on the electrical parameters of the wind farm, photovoltaic power station power generation units and feeders, determine the phase angle difference between the voltage phase angle and the current phase angle, and analyze the change trend of the phase angle difference;

[0023] Perform fluctuation analysis on the curves of active power, reactive power and apparent power in the electrical parameters of the wind farm, photovoltaic power station power generation units and feeders to identify the curve fluctuation characteristics of the power;

[0024] Based on a preset sliding window length, perform variable window length perturbation analysis on the electrical parameters of the wind farm, photovoltaic power station power generation units and feeders to identify the fluctuation characteristics of power perturbation within the window period;

[0025] Based on the change trend of the phase angle difference, the curve fluctuation characteristics of the power and the fluctuation characteristics of the power perturbation, use data statistics methods to extract window feature factors, and use the extracted window feature factors as oscillation risk assessment elements.

[0026] Preferably, based on the window feature factors, through data statistics and trend analysis, and combining the multi-element weight ratio fusion luminescence method for oscillation event recognition, including:

[0027] Based on the extracted window feature factors, determine the elements of the oscillation risk assessment, and assign corresponding weights according to the importance of each element to the oscillation risk assessment;

[0028] Adopt the multi-element weight ratio fusion luminescence method to fuse the results of each element, and display the fusion structure in the form of a numerical value to form a quantitative risk value;

[0029] Judge whether the quantitative risk value exceeds a preset risk threshold. If so, it means that an oscillation event has occurred in the new energy power station; otherwise, it means that no oscillation event has occurred in the new energy power station.

[0030] Preferably, performing data augmentation processing on electrical parameters through a data augmentation-based multimodal fusion strategy, and fusing transfer learning and generative adversarial networks to construct an oscillation pattern recognition model, and using the oscillation pattern recognition model to identify oscillation event patterns, including:

[0031] Use the data transformation method to transform the electrical parameter data and generate preliminary data augmentation samples;

[0032] Based on the preliminary data augmentation samples, and based on circuit laws such as Kirchhoff's law and electromagnetic principles, through this simulated perturbation, generate new electrical parameter data samples;

[0033] Collect external environment data related to the new energy power station, and fuse the new electrical parameter data samples with the external environment data to form a multi-modal data set;

[0034] Use transfer learning technology to transfer the pre-trained convolutional neural network model to the oscillation mode recognition task;

[0035] Construct a generative adversarial network model to generate high-quality simulated oscillation data samples, expand the training data set, use the multi-modal data set and the expanded training data set to train the oscillation mode recognition model, and use the oscillation mode recognition model to identify the oscillation event mode.

[0036] Preferably, the regional master station includes:

[0037] An oscillation data receiving unit for receiving the electrical parameters and oscillation characteristic quantities of the new energy power station sent by the new energy sub-station;

[0038] An oscillation data processing unit for calculating oscillation characteristic quantities according to the electrical parameters of the new energy power station sent by the new energy sub-station, and determining oscillation source discrimination conditions according to the oscillation characteristic quantities;

[0039] A vibration source determination unit for determining the oscillation source in the new energy power station according to the oscillation source auxiliary discrimination conditions;

[0040] An oscillation source action outlet giving unit for judging whether the oscillation source diverges and whether the oscillation duration is greater than a preset time threshold, sorting the comprehensive characteristic values of the new energy power stations participating in the oscillation, sending a generator tripping instruction signal to the new energy sub-station, and calming the oscillation according to the sorting result of the comprehensive characteristic values;

[0041] A generator tripping amount uploading unit for sending the generator tripping amount information to the dispatching main station while sending the generator tripping instruction signal to the new energy sub-station.

[0042] Preferably, the calculating the oscillation characteristic quantities according to the electrical parameters of the new energy power station sent by the new energy sub-station and determining the oscillation source discrimination conditions according to the oscillation characteristic quantities includes:

[0043] An adaptive normalization algorithm based on data entropy normalizes the electrical parameters of a new energy power station to obtain a normalized data set; according to the normalized data set, the amplitude and starting time of each device in the new energy power station are determined through a time analysis window, and the amplitude and starting time of each device in the new energy power station are sorted to obtain the sorting situation of the starting times of each device;

[0044] According to the oscillation mode frequencies in the normalized data set, the oscillation mode frequencies are clustered, and the positive sequence impedance of each oscillation mode is calculated according to the clustering results to obtain the voltage phasor and current phasor at the oscillation mode frequencies; according to the positive sequence impedance characteristic calculation formula, the positive sequence impedance characteristics of the new energy power station equipment in the oscillation mode are calculated;

[0045] Compare the apparent power in the normalized data set with a preset apparent power threshold, and compare the oscillation amplitude in the normalized data set with a preset oscillation amplitude threshold to obtain a comparison result;

[0046] The sorting situation of the starting times of each device, the positive sequence impedance characteristics of the new energy power station equipment in the oscillation mode, and the comparison result are used as auxiliary discrimination conditions for the oscillation source.

[0047] Preferably, the adaptive normalization algorithm based on data entropy normalizes the electrical parameters of the new energy power station to obtain a normalized data set including:

[0048] For each feature quantity in the electrical parameters of the new energy power station, calculate the feature quantity information entropy;

[0049] It should be noted that the higher the entropy value, the greater the degree of data dispersion. The data is adaptively scaled according to the information entropy. For features with high entropy values (large data dispersion), a smaller scaling factor is used; for features with low entropy values (more concentrated data), a larger scaling factor is used.

[0050] Determine the scaling factor according to the feature quantity information entropy, and use the scaling factor to normalize each electrical parameter to obtain a normalized data set;

[0051] Preferably, the clustering process for the oscillation mode frequencies includes:

[0052] Use the Z-score method to standardize the oscillation mode frequency data to obtain standardized oscillation mode frequency data;

[0053] According to the standardized oscillation mode frequency data, calculate the Euclidean distance between each oscillation mode frequency data using the Euclidean distance;

[0054] Take the Euclidean distance between the oscillation mode frequency data as the input of the clustering algorithm, and perform clustering processing on the Euclidean distance between the oscillation mode frequency data through the density peak clustering algorithm to obtain the clustering result;

[0055] Among them, the clustering process of the Euclidean distance between the oscillation mode frequency data through the density peak clustering algorithm includes: constructing a distance matrix according to the Euclidean distance between the oscillation mode frequency data;

[0056] For each oscillation mode frequency data point in the distance matrix, calculate its local density based on the Gaussian kernel function method;

[0057] For each oscillation mode frequency data point, calculate the minimum distance between it and all points with higher density than it;

[0058] Select the clustering centers according to the local density and the minimum distance, and allocate other points to the corresponding clusters according to the density propagation principle to obtain the final clustering result.

[0059] Preferably, the dispatching master station includes: a power system monitoring unit, a generator tripping amount receiving unit, a security checking unit, and an instruction sending unit;

[0060] The power system monitoring unit is used to monitor the power system frequency and voltage level within a preset period;

[0061] The generator tripping amount receiving unit is used to receive the generator tripping amount information sent by the regional master station;

[0062] The security checking unit is used to perform security checking by combining the power system frequency, voltage level, generator tripping amount information and the operation status of the power system. When the quantity and range of the generator tripping amount cause the system frequency and voltage support regulation ability to be exceeded, generate a regional master station locking instruction;

[0063] The instruction sending unit is used to send the generated regional master station locking instruction to the regional master station.

[0064] The beneficial effects of the present invention are:

[0065] 1. The present invention monitors and calculates data such as voltage, current, and oscillation frequency of power generation units or feeders in a wind farm or a photovoltaic power station, calculates oscillation characteristic quantities such as oscillation amplitude, oscillation start time, frequency spectrum, and transient energy in real time, and adds indicators such as voltage active and reactive power sensitivities for the first time to improve the accuracy of oscillation source analysis and positioning. An oscillation online monitoring and analysis function with the ability to receive various oscillation characteristics sent by each sub-station and perform oscillation source positioning is built in the main station in the construction area of the new energy centralized access and collection hub station. It can comprehensively compare according to the oscillation characteristics, accurately identify the oscillation source and key participating power stations within seconds after the oscillation occurs, select the generator tripping according to the oscillation source positioning and send the generator tripping command to the new energy sub-stations. A main station is built in the dispatching center, and system security checks are carried out in combination with system frequency, voltage level, the generator tripping amount of each main station, and the system operation conditions. When problems such as excessive total generator tripping amount and over-wide range occur in a short time, exceeding the system frequency and voltage support and regulation capabilities, a blocking command is sent to the relevant regional main station to achieve effective prevention and control of broadband oscillations on the basis of ensuring the safety of the large power grid.

[0066] 2. In order to effectively cope with the impact of broadband oscillations on the power system, the present invention further designs a collaborative defense mechanism. This mechanism not only relies on the monitoring and prevention and control of individual stations, but also includes the collaborative cooperation between various new energy power stations. By establishing a regional main station to receive and integrate real-time monitoring data from each sub-station, the system can quickly locate the oscillation source, and reasonably dispatch various equipment in the power system according to the characteristics and influence range of the oscillation source, and take coordinated defense measures. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required to be used in the embodiments. Obviously, the following described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts. In the drawings:

[0068] Figure 1 is a schematic diagram of the principle of a broadband oscillation in-situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0069] Figure 2 is a vector diagram of phase angle oscillation in a broadband oscillation in-situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0070] Figure 3 is a vector diagram of amplitude oscillation in a broadband oscillation in-situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0071] Figure 4 is a vector diagram of electrical resonance in a broadband oscillation in-situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0072] Figure 5 It is a waveform feature diagram of phase angle oscillation in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0073] Figure 6 It is a waveform feature diagram of amplitude oscillation in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0074] Figure 7 It is a waveform feature diagram of electrical resonance in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0075] Figure 8 It is a trend classification diagram of oscillation characterization feature elements in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0076] Figure 9 It is a technical path diagram of oscillation sources based on comprehensive index criteria in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0077] Figure 10 It is an auxiliary discrimination flow chart of oscillation energy flow in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0078] Figure 11 It is one of the results of case harmonic source tracing analysis in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0079] Figure 12 It is the second result of case harmonic source tracing analysis in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0080] Figure 13 It is a structural framework diagram of a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0081] Figure 14 It is a communication connection architecture of a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention;

[0082] Figure 15 It is a wide - frequency oscillation mechanism diagram of a new - type power system in a wide - frequency oscillation in - situ monitoring and collaborative defense system according to an embodiment of the present invention.

[0083] In the figure:

[0084] 1. New energy sub - station; 2. Regional main station; 3. Dispatching main station. Specific implementation manner

[0085] To enable those skilled in the art to better understand the technical solutions in this application, the following will clearly and completely describe the technical solutions in the embodiments of this application in conjunction with the accompanying drawings in the embodiments of this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the protection scope of this application.

[0086] According to an embodiment of the present invention, a wide-frequency oscillation in-situ monitoring and collaborative defense system is provided.

[0087] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. As Figure 1 shown, the wide-frequency oscillation in-situ monitoring and collaborative defense system according to the embodiment of the present invention includes:

[0088] A new energy substation 1, configured to obtain the electrical parameters of a new energy power station, calculate the oscillation characteristic quantity of the new energy power station according to the electrical parameters; when the oscillation characteristic quantity reaches a preset oscillation threshold, send the electrical parameters and the oscillation characteristic quantity to the regional master station, and receive and execute the generator tripping instruction issued by the regional master station;

[0089] A regional master station 2, configured to receive the electrical parameters and the oscillation characteristic quantity sent by the new energy substation, and locate the oscillation source by using the wide-frequency oscillation monitoring and tracing technology; select the generators to be tripped according to the oscillation source location result, and send the generator tripping instruction to the new energy substation;

[0090] A dispatching master station 3, configured to monitor the power system frequency, voltage level and the generator tripping quantity information of each regional master station in real time, perform security checking in combination with the power system frequency, voltage level and the generator tripping quantity information of each regional master station, and perform power coordination defense according to the checking result.

[0091] As a preferred embodiment, the new energy substation 1 includes:

[0092] A data acquisition and analysis unit, configured to obtain the electrical parameters of the new energy power station, and identify the oscillation events and oscillation event patterns in the new energy power station by analyzing the development trend of the electrical parameters;

[0093] A regional master station sending unit, configured to send the electrical parameters of the new energy power station to the regional master station after an oscillation occurs in the new energy power station;

[0094] A receiving and executing unit, configured to receive and execute the generator tripping instruction issued by the regional master station.

[0095] As a preferred embodiment, the data acquisition and analysis unit includes:

[0096] The data monitoring unit is used to monitor and analyze the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations in real time by using an oscillation on-line monitoring and analysis device based on the time-division multiplexing method and the adaptive sampling frequency technology. The electrical parameters include voltage, current, active power, reactive power and apparent power.

[0097] The window feature factor extraction unit is used to analyze the voltage-current angle and the power curve fluctuation characteristics according to the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations, and extract the window feature factors by using data statistics and deduction methods.

[0098] The oscillation event recognition unit is used to recognize oscillation events based on the window feature factors through data statistics and trend analysis, and by integrating the multi-factor weight ratio fusion luminescence method.

[0099] The oscillation event pattern recognition unit is used to perform data enhancement processing on electrical parameters through a data-enhanced multi-modal fusion strategy, and fuse transfer learning and generative adversarial networks to construct an oscillation pattern recognition model, and use the oscillation pattern recognition model to recognize the oscillation event pattern.

[0100] As a preferred implementation manner, the method of analyzing the voltage-current angle and the power curve fluctuation characteristics according to the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations, and extracting the window feature factors by using data statistics and deduction methods to form the oscillation risk assessment elements includes: determining the phase angle difference between the voltage phase angle and the current phase angle according to the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations, and analyzing the change trend of the phase angle difference; performing fluctuation analysis on the curves of active power, reactive power and apparent power in the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations to identify the power curve fluctuation characteristics; based on a preset sliding window length, performing variable window length perturbation analysis on the electrical parameters of the power generation units and feeders in wind farms and photovoltaic power stations to identify the fluctuation characteristics of power perturbation within the window period; according to the change trend of the phase angle difference, the power curve fluctuation characteristics and the power perturbation fluctuation characteristics, using data statistics methods to extract the window feature factors, and using the extracted window feature factors as the oscillation risk assessment elements.

[0101] As a preferred implementation manner, the method of recognizing oscillation events based on the window feature factors through data statistics and trend analysis, and by integrating the multi-factor weight ratio fusion luminescence method includes: determining the elements of oscillation risk assessment according to the extracted window feature factors, and assigning corresponding weights according to the importance of each element to the oscillation risk assessment; using the multi-factor weight ratio fusion luminescence method to fuse the results of each element, and presenting the fusion structure in the form of a numerical value to form a quantified risk value; judging whether the quantified risk value exceeds a preset risk threshold. If so, it indicates that an oscillation event has occurred in the new energy power station. Otherwise, it indicates that no oscillation event has occurred in the new energy power station.

[0102] As a preferred implementation, the multimodal fusion strategy through data enhancement performs data enhancement processing on the electrical parameters, and integrates transfer learning and generative adversarial networks to construct an oscillation pattern recognition model. Using the oscillation pattern recognition model to identify the oscillation event pattern includes: using a data transformation method to transform the electrical parameter data to generate preliminary data enhancement samples; based on the preliminary data enhancement samples, based on Kirchhoff's laws and other circuit laws and electromagnetic principles, through this simulated disturbance, new electrical parameter data samples are generated; external environmental data related to the new energy station is collected, and the new electrical parameter data samples are integrated with the external environmental data to form a multimodal data set; using transfer learning technology, the pre-trained convolutional neural network model is migrated to the oscillation pattern recognition task; constructing a generative adversarial network model, generating high-quality simulated oscillation data samples, expanding the training data set, using the multimodal data set and the expanded training data set to train the oscillation pattern recognition model, and using the oscillation pattern recognition model to identify the oscillation event pattern.

[0103] As a preferred embodiment, the regional master station 2 includes:

[0104] An oscillation data receiving unit, used to receive electrical parameters and oscillation characteristic quantities of a new energy station sent by a new energy substation;

[0105] An oscillation data processing unit, used to calculate an oscillation characteristic quantity according to the electrical parameters of the new energy station sent by the new energy substation, and determine an oscillation source discrimination condition according to the oscillation characteristic quantity;

[0106] A vibration source determination unit, used to determine the vibration source in the new energy station according to the vibration source auxiliary discrimination conditions;

[0107] The oscillation source action output giving unit is used to determine whether the oscillation source diverges and whether the oscillation duration is greater than a preset time threshold, and to sort the comprehensive characteristic values ​​of the new energy stations participating in the oscillation, send a power-off command signal to the new energy substation, and calm the oscillation according to the comprehensive characteristic value sorting result;

[0108] The machine cutting quantity uploading unit is used to send the machine cutting quantity information to the dispatching main station at the same time when sending the machine cutting instruction signal to the new energy substation.

[0109] As a preferred embodiment, calculating the oscillation characteristic quantity based on the electrical parameters of the new energy power station sent by the new energy substation, and determining the oscillation source discrimination condition according to the oscillation characteristic quantity includes: using the adaptive normalization algorithm based on data entropy to normalize the electrical parameters of the new energy power station to obtain a normalized data set; according to the normalized data set, determining the amplitude and starting oscillation time of each device in the new energy power station through a time analysis window, and sorting the amplitude and starting oscillation time of each device in the new energy power station to obtain the sorting situation of the starting oscillation time of each device; according to the oscillation mode frequency in the normalized data set, performing clustering processing on the oscillation mode frequency, and calculating the positive sequence impedance of each oscillation mode according to the clustering result to obtain the voltage phasor and current phasor at the oscillation mode frequency; calculating the positive sequence impedance characteristic of the new energy power station device in the oscillation mode according to the positive sequence impedance characteristic calculation formula; comparing the apparent power in the normalized data set with a preset apparent power threshold, and comparing the oscillation amplitude in the normalized data set with a preset oscillation amplitude threshold to obtain a comparison result; using the sorting situation of the starting oscillation time of each device, the positive sequence impedance characteristic of the new energy power station device in the oscillation mode, and the comparison result as the oscillation source auxiliary discrimination condition.

[0110] The positive sequence impedance characteristic calculation formula is:

[0111]

[0112] In the formula, represents the positive sequence impedance in each oscillation mode, represents the voltage phasor, represents the current phasor, j represents the imaginary part, R m represents the resistance in each oscillation mode, X m represents the reactance in each oscillation mode.

[0113] As a preferred embodiment, the adaptive normalization algorithm based on data entropy normalizes the electrical parameters of the new energy power station to obtain a normalized data set, including: for each characteristic quantity in the electrical parameters of the new energy power station, calculating the characteristic quantity information entropy; determining a scaling factor according to the characteristic quantity information entropy, and using the scaling factor to normalize each electrical parameter to obtain a normalized data set;

[0114] The calculation formula of the characteristic quantity information entropy is:

[0115]

[0116] In the formula, H(x) represents the information entropy of the characteristic quantity x, p(x i) represents the probability of the occurrence of the i-th feature quantity x, and n represents the number of types of feature quantities. The higher the entropy value, the greater the degree of data dispersion. The data is adaptively scaled according to the information entropy. For features with a high entropy value (large data dispersion), a smaller scaling factor is used; for features with a low entropy value (relatively concentrated data), a larger scaling factor is used.

[0117] The calculation formula of the scaling factor is as follows:

[0118]

[0119] In the formula, X norm represents the scaling factor, x represents the feature quantity in the electrical parameters, min(x) represents the minimum value of the feature quantity in the electrical parameters, max(x) represents the maximum value of the feature quantity in the electrical parameters, H(x) represents the information entropy of the feature quantity x, and e represents a constant. e is an extremely small constant (such as 10 -8 ), to prevent the denominator from being zero. This can consider the discrete characteristics of the data during normalization, avoid data with large dispersion from dominating the normalization result, better retain the differences of different features, reduce data redundancy, and can also perform normalization according to the internal structure of the data, enhancing the distinguishability of the data.

[0120] As a preferred embodiment, the clustering process for the oscillation mode frequencies includes:

[0121] Using the Z-score method to standardize the oscillation mode frequency data to obtain standardized oscillation mode frequency data; according to the standardized oscillation mode frequency data, calculating the Euclidean distance between each oscillation mode frequency data using the Euclidean distance; taking the Euclidean distance between each oscillation mode frequency data as the input of the clustering algorithm, and performing clustering processing on the Euclidean distance between each oscillation mode frequency data through the density peak clustering algorithm to obtain a clustering result;

[0122] Among them, the clustering process for the Euclidean distance between each oscillation mode frequency data through the density peak clustering algorithm includes: constructing a distance matrix based on the Euclidean distance between each oscillation mode frequency data; for each oscillation mode frequency data point in the distance matrix, calculating its local density based on the Gaussian kernel function method; for each oscillation mode frequency data point, calculating the minimum distance between it and all points with a higher density than it; selecting cluster centers according to the local density and the minimum distance, and allocating other points to the corresponding clusters according to the density propagation principle to obtain the final clustering result.

[0123] As a preferred embodiment, the dispatching master station 3 includes: a power system monitoring unit, a generator tripping amount receiving unit, a security checking unit, and an instruction sending unit;

[0124] A power system monitoring unit, which is used to monitor the frequency and voltage level of the power system within a preset period;

[0125] A generator tripping amount receiving unit, which is used to receive the generator tripping amount information sent by the regional master station 2;

[0126] A security check unit, which is used to perform security checks by combining the power system frequency, voltage level, generator tripping amount information and the operation conditions of the power system. When the quantity and scope of the generator tripping amount lead to exceeding the system frequency and voltage support regulation capabilities, a regional master station locking instruction is generated;

[0127] An instruction sending unit, which is used to send the generated regional master station locking instruction to the regional master station 2.

[0128] To facilitate the understanding of the above technical solution of the present invention, the above technical solution of the present invention will be further described from the perspectives of architecture and principle as follows:

[0129] 1. Main technical route;

[0130] The oscillation mechanisms of the new power system are diverse. As Figure 15 shown, and the coupling and correlation are relatively complex. The affected frequency domain covers hundreds or even thousands of hertz, and the oscillation patterns of the power curves can no longer be clearly observed. It is technically difficult to clarify the inducement of oscillation and predict oscillation from the perspective of traditional mechanisms. Since the three core elements for evaluating the stability level of key electrical quantities in the power system are "frequency", "waveform", and "amplitude", and the main impacts of oscillation also focus on the damage and distortion of the above core elements of electrical quantities. Therefore, starting from the mutations of the three elements, combining the different characteristics of the time domain and frequency domain, the characteristics and trends of electrical quantities during the occurrence of oscillation can be analyzed to warn and locate the occurrence of oscillation. When measuring the frequency, the frequency is generally associated with the vector, and the specific relationship formula is shown as follows:

[0131]

[0132] In the formula, represents the prediction of oscillation, Δf represents the frequency difference, f 0 represents the initial frequency, k represents the index of the current sampling point, N represents the modulation coefficient, r represents the index of the subcarrier, represents the phase shift of the rth signal; X represents the signal amplitude.

[0133] Among them, as shown in Table 1, it is Figure 15 the annotation of Chinese and English phrases;

[0134] Table 1 Annotation of Chinese and English phrases

[0135]

[0136]

[0137] The current electrical quantity frequency is calculated based on the voltage waveform. In the processing of power system sampling signals, it is assumed that the change in frequency within one cycle time is very small. Therefore, the frequency difference Δf(t) within the data window time range is expressed as a fixed value Δf. Among them, in the assumption of the model, it is also based on the constant amplitude within a short time, so as to obtain a phase angle and frequency relationship independent of the amplitude as shown in the following formula:

[0138]

[0139] In the formula, θ represents the phase angle of the power system at time t, Δf represents the frequency difference; f 0 represents the initial frequency; N represents the modulation coefficient.

[0140] According to the above phase angle formula, it can be deduced that the capture of frequency elements before oscillation occurs can start from the analysis of the sudden change of the phase angle. The frequency oscillation can also evolve into a phase angle oscillation. At the same time, combined with the periodic change of the amplitude, the oscillation phenomenon can be classified as: phase angle oscillation, amplitude oscillation and electrical resonance. Among them, the vector diagrams of phase angle oscillation, amplitude oscillation and electrical resonance are as Figures 2 - 4 shown, and the waveform characteristics of phase angle oscillation, amplitude oscillation and electrical resonance are as Figures 5 - 7 shown. The expressions of phase angle oscillation, amplitude oscillation and electrical resonance are respectively:

[0141]

[0142]

[0143] In the formula, Y p represents the electrical vector after phase angle oscillation, Y m represents the electrical vector after amplitude oscillation, Y c represents the electrical vector after electrical resonance, y 0 represents the electrical vector before oscillation occurs, j represents the imaginary part, ω 0 represents the initial angular frequency, represents the initial phase angle, y R represents the change amount of the electrical vector after oscillation occurs, α represents the attenuation factor of the oscillation signal, t represents time, represents the phase angle of y R ω R represents the angular frequency of y R R

[0144] The main forms of oscillation can be roughly classified into changes in phase angle and amplitude, and can be reflected by two quantities that can be intuitively felt at the message data level. The envelope of the power curve of this oscillation changes significantly, and the characteristics of electrical quantities before oscillation can be clearly defined and analyzed. According to the severity and sequence of changes in oscillation characteristic quantities, the oscillation phenomena can be classified into 9 main directions, such as Figure 8 shown Figure 8 in which, ① low-frequency oscillation of synchronous machines, ② subsynchronous oscillation dominated by the shaft system of synchronous machines, ③ converter PLL / power synchronization oscillation, ④ stability problems dominated by the outer loop of converters, ⑤ stability problems dominated by the outer loop of conventional direct current, ⑥ stability problems induced by the slip of asynchronous machines, ⑦ subsynchronous oscillation caused by adding series compensation to synchronous machines, ⑧ subsynchronous oscillation caused by adding series compensation to doubly-fed units, ⑨ medium- and high-frequency oscillation dominated by the inner loop / feedforward of converters.

[0145] As Figure 8 shown, for the oscillation early warning of thermal power units, it can start from the trend deviation of the phase angle. For subsynchronous oscillation, etc., it is necessary to superimpose the trend increase of the oscillation amplitude. For example, in classifications ①②③, the trend growth of the phase angle will receive key attention. For those involving new energy such as ④⑤⑥, it is more reflected in the trend change of the amplitude. This characteristic differentiation in trend can become the key basis for trend discrimination. Therefore, the main idea of early warning positioning is to find key characteristic quantities on the basis of mechanism analysis, and analyze the development trend of characteristic quantities in the time domain, the distribution trend in the frequency domain, etc., as Figure 9 shown Figure 9 in which, u1, u2, un represent electrical quantities such as voltage, current, and power, y1, y2, yn represent the characteristic values of electrical quantities obtained through frequency domain and time domain calculations for oscillation analysis and positioning, and z is the result after comprehensive criterion fusion calculation.

[0146] As Figure 9 shown, the basic concept of early warning positioning is based on the frequency analysis of disturbance components, constructing a variable window length disturbance energy integration algorithm, calculating the growth trend of disturbance energy, combining the change trends of the voltage-current angle, port impedance, and the fluctuation characteristics of the power curve, etc., and comprehensively analyzing and comparing the characteristic quantities of each substation on the regional master station side, accurately positioning the oscillation source and issuing control instructions, and conducting a whole-network security check on the dispatching master station to form a wide-area wide-frequency oscillation early warning, positioning, and defense system for large power grids based on oscillation path deduction and fixed threshold determination.

[0147] 2. On-site oscillation monitoring and eigenvalue calculation method for substations;

[0148] The main function of on-site monitoring at the site is to conduct real-time monitoring and identification of oscillation modes by analyzing the development trend of electrical characteristic quantities in the time domain and the distribution in the frequency domain after oscillation occurs. When oscillation occurs in the monitoring equipment, an alarm is triggered and recording is started. The frequency, amplitude, time and other oscillation characteristic quantities of the monitored oscillation are transmitted to the regional master station.

[0149] In order to collect and transmit a large number of broadband electrical quantities in wind farms, photovoltaic power station power generation units and feeders, time division multiplexing is used to achieve synchronous collection to ensure the time consistency of data. In terms of collection frequency, according to the changing characteristics of different electrical parameters, adaptive sampling frequency technology is used. For parameters such as current and voltage that change rapidly, the sampling frequency is automatically increased to more than 10kHz when oscillation occurs, and the sampling frequency is reduced to 1kHz during steady-state operation to reduce the amount of data and processing burden.

[0150] According to the analysis and research of historical operation and oscillation data, innovative window feature identification of electrical quantities before oscillation is carried out. The main methods of data statistics and deduction are used to extract window feature factors for changes in disturbance power density, harmonic power growth trend, harmonic frequency distribution change, voltage and current phase angle change trend, etc., to form oscillation risk assessment factors, and use weight matching to form an overall oscillation early warning strategy that integrates all factors. The window period refers to a period before the oscillation power is significantly enhanced, which will include abnormal symptomatic features such as power disturbances and harmonic frequency points. As shown in the figure, based on the analysis of actual case data, it can be clearly observed that during the window period before the oscillation, the time window of drastic impedance changes highly overlaps with the change in the voltage-current angle, and is positively correlated with the intensity of the disturbance power. Before the angle trend changes, the impedance parameters have abnormally fluctuated significantly.

[0151] The present invention innovatively adopts a risk assessment method based on electrical quantity signs in the pre-oscillation window period, and the specific steps are as follows:

[0152] 1) Pre-oscillation window feature separation and trend analysis function: Taking the primary equipment corresponding to the broadband data acquisition as the node, the broadband substation data is analyzed in a normalized manner. Through historical data and online data analysis, the disturbance energy trend, impedance and other characteristic quantities of each node are monitored in real time. The number of disturbances occurring per unit time, harmonic power, main harmonic frequency and other information are displayed in real time, and the corresponding curves are drawn to identify the multi-band disturbance intensity and distribution characteristics of the monitoring node, and form a baseline to determine the abnormal working conditions of the monitoring node.

[0153] 2) Oscillation risk quantification and early warning function: During the normal data statistics process, according to the changes in disturbance power density, the growth trend of harmonic power, and the changes in harmonic frequency distribution, a trend chart is constructed. A stable trend indicates a relatively low current risk of the power grid. Conversely, if the trend changes of each monitoring element exceed the threshold, the oscillation risk assessment logic is triggered. During the triggered window period, the trends of each element are comprehensively judged. The multi-element weight ratio fusion luminescence method is adopted to display the comprehensive results of each element in the form of numerical values and draw them into a data curve. The trend of the curve represents the intensity of the disturbance risk. By comparing with the characteristic baselines of each monitoring node, an early warning of the deviation trend is formed, and thus an oscillation risk assessment early warning is formed. If each element returns to stability again, the risk assessment is exited and the normal monitoring is resumed.

[0154] 3) Input and output description: The input data includes (1) the real-time and historical data of the monitoring nodes, which can be the mixed data of PMU and broadband monitoring devices, not less than 10 days. (2) The online data of the monitoring nodes: the fundamental wave power values collected by the PMU or broadband substation, the top 5 main harmonic frequency points and power values, voltage / current frequencies and amplitudes, impedance, phase angle, etc. The output data includes: (1) the change curves of window characteristic factors such as the change in disturbance power density, the growth trend of harmonic power, the change in harmonic frequency distribution, the change trend of impedance and voltage current phase angle, etc.; (2) the quantitative risk numerical curve.

[0155] The present invention optimizes the training process by using an innovative algorithm model enhanced by fusion transfer learning and generative adversarial network (GAN): 1) Transfer learning is introduced in the training of a multi-layer convolutional neural network (CNN). Given the high cost of obtaining power system oscillation data, a CNN model is pre-trained from a large number of labeled data in offline electromagnetic transient simulations to learn the general time-frequency domain feature extraction mode. Then, the pre-trained model is transferred to the power system broadband oscillation monitoring task, and the model is fine-tuned using a small amount of local power system data. This method can accelerate the convergence of the model and reduce the dependence on large-scale power system data. 2) Enhancing the training data based on generative adversarial network (GAN): To address the problem of scarce data for some rare oscillation patterns in power system oscillation data, a GAN model is constructed. The generator learns the distribution law of the existing oscillation data and generates simulated oscillation data samples; the discriminator distinguishes between real data and generated data. During the training process, the generator and the discriminator compete with each other and optimize collaboratively. The generated high-quality simulated data is mixed with the real data to expand the training data set. This not only increases the diversity of the data but also enables the model to learn richer oscillation characteristics and improves the recognition and prediction ability for rare oscillation patterns.

[0156] Improving the training data cleaning and processing mode to enhance the model accuracy:

[0157] Multimodal Fusion Strategy with Data Augmentation: Perform multimodal data augmentation on the collected electrical parameter data (such as voltage, current, power, etc.). In addition to traditional data transformation methods (such as translation, scaling, rotation), combined with the physical characteristics of the power system, according to circuit laws and electromagnetic principles, simulate perturbations on the data. Based on Kirchhoff's laws, simulate the changes in electrical parameters when the parameters of a certain component in the circuit change, and generate new training data. Integrate the electrical parameter data with external environment data (such as temperature, humidity, wind speed, etc.) and use it as multimodal data for model training. External environmental factors may affect the performance of power equipment and are thus related to the occurrence of oscillations. By fusing multimodal data, the model can learn more comprehensive features and improve the accuracy of oscillation prediction.

[0158] Specifically, in the frequency domain, mainly based on the frequency analysis of perturbation components and the variable window length perturbation energy integration algorithm, analyze the growth trend of perturbation energy. For oscillation events in different frequency bands, use a variety of spectrum analysis algorithms (such as FFT, PRONY, transient energy algorithm) to analyze the time-frequency characteristics of oscillations in the range of 0.1 - 2500 Hz, and combine the spectrum analysis of electrical quantities to calculate the broadband impedance characteristics, analyze the change trend of port impedance, and assist in locating the oscillation source based on the impedance characteristics of the equipment. Integrate the perturbation energy flow to form oscillation patterns and monitor and warn about the impedance of converters. In the time domain, carry out variable window length perturbation analysis and calculation of voltage, current, and power, analyze the characteristics of the voltage-current angle and power curve fluctuations, add dynamic electrical characteristic quantities related to oscillations on the basis of traditional oscillation amplitude and frequency, and utilize the characteristics that the equipment voltage fluctuates greatly with active and reactive power before the occurrence of broadband oscillations to add true labels of dynamic characteristic quantities such as voltage active sensitivity dv / dp and voltage reactive sensitivity dv / dq to assist in warning and locating the oscillation source. Through the analysis of the data of the on-line monitoring system, calculate the key characteristic quantities of broadband oscillations in a certain frequency band, and then conduct comprehensive analysis. The specific calculation is as follows:

[0159] (1) Oscillation amplitudes of active power P, reactive power Q, and apparent power S: Perform spectrum analysis on the power values of P, Q, and S to calculate the instantaneous power amplitudes P max , Q max , S max . The oscillation monitoring start criterion is that the instantaneous power oscillation amplitude exceeds the preset thresholds P WFO , Q WFO , S WFO and lasts for M seconds. Record the above oscillation amplitudes and the oscillation start time T WFO , P WFO . The values of P

[0160] (2) Oscillation displacement integration of P, Q, and S: Perform spectral analysis on the P, Q, and S power values. For the monitored key oscillation frequency bands, calculate the oscillation displacements ∫|p(t)|dt, ∫|q(t)|dt, and ∫|s(t)|dt corresponding to the time window t. If the oscillation is an equal-amplitude oscillation, this integral quantity is a constant value within a fixed time window; otherwise, it is a damped or amplified oscillation. This indicator mainly depicts the intensity of rapidly changing disturbance energy from the perspective of sorting and statistics. Under steady-state operation, the average value of the power curve in each observation window shows a stable trend. With the emergence of gradually increasing high-frequency disturbance energy, the power curve fluctuates significantly. Before an observable typical periodic oscillation waveform appears, the average value of the power often shows many crossings. This criterion is very suitable for medium-high frequency or scenarios with relatively large oscillation energy.

[0161] In the formula, p(t) represents the instantaneous active power, q(t) represents the instantaneous reactive power, s(t) represents the instantaneous apparent power, and P, Q, and S represent the active power, reactive power, and apparent power respectively.

[0162] (3) Oscillation disturbance energy distribution: For the oscillation displacement p(θ) corresponding to the phase angle θ of each power generation element in the new energy power station, calculate ∫p(θ)dθ within a fixed window to construct the quantification of the disturbance magnitude in the observation window. This indicator mainly depicts the development trend of the oscillation power curve. By comprehensively comparing the oscillation disturbance energies of multiple devices, the oscillation occurrence path can be deduced.

[0163] In the formula, θ is the phase angle of each power generation element, and p(θ) is the oscillation displacement corresponding to the phase angle θ of the power generation element.

[0164] (4) Oscillation amplitude of the power factor angle: Perform spectral analysis on the value to obtain the fluctuation amplitude of the power factor angle Power factor characteristics: The power factor reflects the development trend of the phase angle difference between voltage and current. Under steady-state operation, the phase angle difference between voltage and current should fluctuate within a certain range. When there is an oscillation risk or oscillation occurs, the phase angle difference will show a spreading characteristic. By analyzing the power factor angle, the oscillation disturbance risk of the warning device can be extracted. This indicator mainly depicts the development trend of the phase angle difference between voltage and current. Under steady-state operation, the phase angle difference between voltage and current should fluctuate within a certain range. However, in the early stage of oscillation, the phase angle difference will show a spreading characteristic. The trend of the phase angle difference is also one of the main weighting factors for oscillation localization.

[0165] (5) Oscillation amplitudes of voltage U and current I: Perform spectral analysis on the U and I values to obtain the oscillation amplitudes ΔU and ΔI under the current key oscillation frequency band. In the case of pure active or pure reactive oscillations that may occur in the new energy power station, ∫|p(t)|dt, ∫|q(t)|dt, ∫|s(t)|dt, ∫p(θ)dθ, The criterion fails, and it is necessary to comprehensively consider the collector line, SVG, synchronous condenser, voltage and current oscillation amplitudes of wind turbines as an auxiliary basis for oscillation source location. During steady-state operation, there will be a certain proportion of disturbance components in the voltage and current. In the early stage of oscillation, the proportion often shows an upward trend. Therefore, the analysis of the disturbance trend of voltage and current is one of the main weighting factors for oscillation source location.

[0166] (6) Voltage active sensitivity, voltage reactive sensitivity: When the new energy converter operates near the critical point of the PV curve and the voltage active and voltage reactive sensitivities are relatively large, broadband oscillation is likely to occur. Based on this characteristic, the electrical characteristic quantities of dv / dp and dv / dq under a fixed time window are used as oscillation early warning discrimination indicators. At the same time, under different converter control parameters and structures, there are also certain differences reflected in dv / dp and dv / dq. Then this characteristic represents a certain converter impedance characteristic, and it can reflect whether the device is an oscillation source from the impedance perspective.

[0167] (7) Voltage and current harmonic components: Through spectrum analysis, obtain the proportion and spectrum distribution of non-power frequency components in voltage and current. There is an oscillation risk in the converter impedance, and there may be a small amount of non-power frequency disturbance components, which show an upward trend. Based on this characteristic, predict the impedance model.

[0168] (8) After the above indicators are calculated, send the above indicators to the regional master station, and the regional master station will conduct oscillation source discrimination and measure selection.

[0169] (9) Communication, time synchronization and recording function;

[0170] 1) Support the GB / T26865.2 standard, and have the function of transmitting synchronous phasors, interharmonics, harmonics, alarm information and recording files externally; support the DL / T860 standard, and have the function of transmitting alarm information and status events externally; it can also have the related function of transmitting fundamental wave, harmonic and interharmonic measurement data externally;

[0171] 2) It should have a timekeeping function. When the synchronous time signal is lost or abnormal, the timekeeping accuracy of the device is within 60 minutes, and the change in the phase angle measurement error is not greater than 1°;

[0172] 3) Support two methods of triggered recording and continuous recording; Triggered recording supports methods such as alarm, manual and network triggering; The sampling rate of triggered recording should not be lower than 12.8 kHz. When the alarm signal persists, the recording time is not less than 60 s, and the number of recording files is not less than 256, with cyclic recording.

[0173] Specifically, when clustering the oscillation mode frequency data, by introducing the Density Peaks Clustering (DPC) algorithm, a more efficient and accurate clustering effect is achieved, enhancing the uniqueness of the solution. The optimization of the Density Peaks Clustering (DPC) algorithm for clustering steps is as follows: After calculating based on the Euclidean distance, the obtained distance matrix is used as the input for the density peaks clustering algorithm. The DPC algorithm identifies the clustering centers based on two key factors: the local density of the data points and the distance to the high-density points. For each oscillation mode frequency data point, its local density is calculated. The calculation of the local density can adopt a method based on the Gaussian kernel function, calculating the minimum distance between each point and the points with higher density than it. After determining the clustering centers, according to the principle of density propagation, other points are assigned to the corresponding clusters. Compared with traditional clustering algorithms, the DPC algorithm does not require pre-setting the number of clusters, can more accurately discover the natural clustering structure in the dataset, and can achieve a more reasonable clustering division for the complex and irregular distribution in the oscillation mode frequency data, thus providing a more reliable basis for the subsequent positioning and analysis of the oscillation source.

[0174] 3. Oscillation source location by the regional master station based on comprehensive criteria;

[0175] (1) Oscillation source location and coordinated defense system based on comprehensive criteria;

[0176] In a power grid with a high proportion of new energy, wide-frequency oscillation events are often the result of the interaction of nearby devices. The devices participating in the oscillation are distributed within a certain range near the collection station, and there are multiple wide-frequency monitoring devices in the electrical strong coupling area that can monitor the oscillation events. However, relying on local monitoring devices cannot accurately determine whether the oscillation is only the oscillation of this device, how large the oscillation range is, and which devices contribute greatly to the oscillation, making it difficult to make effective control decisions. Especially for the centralized access of new energy clusters in desert, gobi, and wasteland areas, large-scale new energy oscillations may occur. Simply and crudely taking local cut-off measures will lead to large-scale disconnection of new energy, having a greater impact on the safety of the large power grid. Therefore, it is necessary to comprehensively collect the oscillation event index information of the lower-level new energy power stations near the collection station to determine the oscillation range, oscillation mode, and the relative magnitude of the participation degree of each device or node in the oscillation, and make correct control decisions on the specific unit output or feeder according to this information.

[0177] The broadband oscillation in-situ monitoring and collaborative defense solution proposed by the present invention can timely obtain complete information on broadband oscillation events with a large affected range in the region, including detailed oscillation analysis results such as the oscillation affected range, the clustering of oscillating units, the oscillation interface, and the relative contribution degree of each device to the oscillation. For forced oscillation, the oscillation source is located according to the relative magnitude of the transient energy flow output by new energy units or SVG, or the oscillation outflow region is located according to the transient energy flow direction of the power grid topology. By comparing and analyzing the key characteristic indicators of broadband oscillation in new energy stations in the region, comprehensive analysis and monitoring of the oscillation situation are realized, including online analysis of oscillation modes, unit clustering, source location, and action outlets, supporting the comprehensive monitoring and prevention and control of cross-station broadband oscillation events.

[0178] (2) Oscillation source location based on the comprehensive criterion of oscillation characteristic quantities;

[0179] After the regional master station receives the oscillation characteristic quantities calculated by the sub-stations, in the oscillation source algorithm, the power oscillation amplitude and oscillation energy are used as the main judgment basis, and various oscillation source determination indicators are comprehensively considered. The summary table of the determination indicator weights is shown in Table 2:

[0180] Table 2 Summary Table of Oscillation Source Determination Indicator Weights

[0181]

[0182]

[0183] Among them, the specific calculation description of the oscillation source location is as follows:

[0184] 1) The master station performs comprehensive collaborative defense control by receiving index information such as the start time of the oscillation event (emphasizing the actual impact of the oscillation on the system or the time point when the system response is caused), the duration, the modal frequency, the alarm device, the oscillation amplitude, and the maximum and minimum values of the modal voltage / current sent by each sub-station. For the convenience of comprehensive comparison of devices with different models and capacities, normalized calculations are used in the calculation of each index.

[0185] 2) Amplitude and starting oscillation time criterion: Sort the amplitudes and starting oscillation times (the time when the oscillation criterion is started, that is, the initial moment when the system enters the oscillation state from the stable state) of each device in a specific time analysis window, and use the sorting situation of the starting oscillation times of each device as an auxiliary discrimination condition for the oscillation source.

[0186] The amplitude comparison of each device uses the Min-Max Scaling algorithm to retain the original distribution characteristics of the data and better reflect the relative magnitude of the amplitudes of each device. After the above processing, the amplitude data will be scaled to the range of 0 to 1.

[0187] In the analysis of the starting oscillation time, the main concern is the relative order of each device. At this time, the purpose of normalization is to convert the data into a unified scale for easy comparison. The Rank Normalization method is adopted. The data is arranged in ascending order, and each data point is assigned a rank. Then, the rank is divided by the total number of data points plus one (if the total number of data points is n, then each rank is divided by n + 1). The new values obtained in this way will be between 0 and 1.

[0188] 3) Impedance criterion:

[0189] Monitor the amplitudes, phase angles, and frequencies of the harmonic voltages and currents among the three phases of the monitoring device. Through data spectrum analysis, calculate the frequency and amplitude data of the inter-harmonic oscillation; cluster the modal frequencies, calculate the positive-sequence impedance of each mode according to the clustering results, and obtain the voltage phasor at the oscillation modal frequency and current phasor Calculate the positive-sequence impedance characteristic of the converter in this mode. When R m , X m exceeds the given threshold, an alarm is issued.

[0190] When R m <0 indicates that this component is the active power source for the oscillation of this mode;

[0191] When X m <0 indicates that this component is the reactive power source for the oscillation of this mode.

[0192] Since the phase calculation requires the Prony algorithm and there is a certain delay, it does not have real-time online performance and is only used as an auxiliary judgment. The device / station with the smallest absolute value of negative impedance is given in rounds. The calculation formula for the positive-sequence impedance characteristic is:

[0193]

[0194] In the formula, represents the positive-sequence impedance under each oscillation mode, represents the voltage phasor, represents the current phasor, J represents the imaginary part, R m represents the resistance under each oscillation mode, X m represents the reactance under each oscillation mode.

[0195] Among them, the steps for clustering the modal frequencies are:

[0196] Step 1. To avoid the influence of data with different dimensions, it is necessary to perform standardization processing on the frequency data using the Z-score method.

[0197]

[0198] In the formula, xi represents the i-th frequency value; μ represents the mean value of the frequency; σ represents the standard deviation;

[0199] Step 2: Use Euclidean distance to calculate the distance between each modal frequency as the input of the clustering algorithm;

[0200]

[0201] Where d(x, y) represents the Euclidean distance between the modal frequencies, and x and y represent the eigenvectors of the modal frequencies;

[0202] Step 3: Select a suitable clustering method based on the data distribution characteristics and actual needs, and input the preprocessed modal frequency data into the selected clustering algorithm. According to the clustering results, analyze the modal frequencies in each cluster to confirm whether they meet the oscillation mode characteristics.

[0203] 4) Condition 1 for the device to participate in the oscillation source calculation: the apparent power is greater than a given threshold; Condition 2 for the device to participate in the oscillation source calculation: the oscillation amplitude is greater than a given threshold;

[0204] 5) Provide the oscillation source judgment mark condition: if the oscillation amplitude of the device is the largest and the ratio of the amplitude to the second lowest is greater than the given threshold, the oscillation source alarm action is initiated;

[0205] 6) Give the exit conditions for the oscillation source action: when the oscillation source diverges (the monitoring and acquisition of the oscillation source divergence depends on the in-depth analysis of the changes in the oscillation characteristic quantities, especially the evolution of the oscillation amplitude and frequency. The system uses an algorithm to determine the changing trend of the oscillation mode and identify whether there is divergence behavior), the oscillation duration (the time window in which the oscillation source is detected and persists in the system) is greater than the given threshold TWFO, the oscillation source alarm action is initiated, the oscillation source comprehensive criterion weight is ranked high, and the difference from the second lowest value is greater than a certain threshold, and an action exit signal is sent to the substation; the comprehensive characteristic values ​​of the new energy stations participating in the oscillation are sorted, and the station with the largest ranking is removed in the first round; if the oscillation does not subside, the station (or multiple stations) with the largest comprehensive characteristic value at that time will be removed in the second round until the oscillation subsides.

[0206] Among them, the comprehensive characteristic value of the new energy station may be obtained by weighted summary calculation of multiple monitoring indicators (such as voltage, current, frequency, oscillation amplitude, etc.). Each characteristic value may represent a certain aspect of the oscillation characteristics, while the comprehensive characteristic value uses a certain weighting method to integrate the influence of multiple indicators to form an overall evaluation.

[0207] 7) According to the oscillation energy reported by each substation, the oscillation source is identified based on the oscillation energy flow. Since the calculation of the oscillation energy flow requires a certain delay, this criterion is also used as an auxiliary criterion and can be used for oscillation source analysis. The specific analysis process decision tree is as follows:Figure 10 The details are as follows:

[0208] Calculate the oscillation mode and transient energy integral value of the unit output and plant injection power of the synchronous power grid, and analyze the amplitude and energy value:

[0209] When only one unit (or multiple units in the same power plant) meets the oscillation alarm threshold, and the transient energy flow value output by the unit (or the power plant) is the largest, it is judged as a forced oscillation, and the unit (or the power plant) is the source of forced oscillation;

[0210] When multiple units meet the oscillation alarm threshold:

[0211] If the transient energy flow value output by one of the units (or multiple units in the same power plant) is the largest, and the transient energy flow values ​​output by other units are relatively small, or negative, it is judged to be a forced oscillation combined with unit resonance, and the unit (or the power plant) is the source of forced oscillation;

[0212] If there is no power plant with obvious transient energy flow output, but the transient energy flow value output by a certain substation is the largest, and the transient energy flow values ​​of other units are relatively small, or negative, it is judged to be forced oscillation combined with unit resonance, and the oscillation source unit is in the lower-level power grid of the substation;

[0213] If the transient energy integral of the power injected by the unit and the substation has no obvious differentiation characteristics and the oscillation source is not found, it is judged to be a weakly damped oscillation.

[0214] Oscillating energy flow is mainly used to locate the source of forced oscillation. Most of the oscillations in the power grid are forced oscillations with clear external disturbance sources. When the frequency of the external forced oscillation source happens to match the inherent oscillation mode of the power grid, it may trigger resonance or resonance phenomenon, resulting in unit grouping and large-scale power fluctuations. The resonance phenomenon at this time may cause the power grid to continue to oscillate and has similarities with the characteristics of weak damping oscillation. When analyzing and distinguishing the oscillation mode of the power grid, a reasonable strategy is to first determine whether there is an external forced oscillation source and take corresponding measures by determining the oscillation source. And through the analysis of the energy flow of the power grid, using the topological structure and energy transmission characteristics of the power grid, through the energy flow and changes in the power grid, it is possible to trace back to the specific divergence area of ​​the oscillation energy, such as a regional power grid or a transformer, which can also be identified as forced oscillation. When multiple units oscillate in the same mode and the power grid has large-scale power oscillations, and no specific oscillation source or oscillation area is located, it can be considered that a weak damping oscillation has occurred.

[0215] (3) Typical case data analysis, such as Figures 11 - 12 The figure shows a typical wind power collection area power grid schematic diagram.

[0216] Harmonic source tracing analysis is carried out based on the continuous recorded wave data of broadband measurement devices at all nodes during on-site subsynchronous oscillation: for the original sampled values of three-phase voltages and three-phase currents at each node, information such as the dominant modal frequency, amplitude, phase angle, and oscillation start time is obtained through feature extraction and modal clustering, and characteristic quantities such as modal impedance, complex power, and oscillation energy flow at each node are calculated. According to the sorting of the comprehensive oscillation characteristic values, the comprehensive oscillation characteristic value of the Naomao Lake Wind Farm is the largest. As shown by the arrows in Figure 11 , the modal power flow flows from the wind farm to the power grid, which also determines that the oscillation source is the wind farm, thus verifying the effectiveness and accuracy of the proposed comprehensive criterion for broadband oscillation sources.

[0217] 4. Dispatching master station security checking method;

[0218] Integrate the system frequency, voltage level, the amount of generator tripping at each master station, and the system operation conditions to carry out system security checking. When problems such as excessive total generator tripping amount and wide range occur in a short time, resulting in exceeding the system frequency and voltage support regulation capabilities, such as the maximum power imbalance in the Northwest Power Grid being 2.1 million kilowatts, and the amount of generator tripping issued by each regional master station exceeding 2.1 million kilowatts within a specific time interval, a locking instruction for a certain period is sent to the relevant regional master stations. After the frequency recovers, the locking instruction is stopped, and the regional master stations continue to implement oscillation prevention and control to effectively prevent and control broadband oscillation on the basis of ensuring the safety of the large power grid.

[0219] 5. The system functional architecture is as follows:

[0220] The overall control system is divided into three major parts: the dispatching general station, the regional master stations, and the new energy sub-stations. As shown in Figures 13 - 14 , the functions of each part are briefly described as follows:

[0221] Dispatching general station: 1) Deployed at the dispatching end, it can monitor the system frequency, voltage level within a certain period in real time, and the information of the amount of generator tripping sent by each regional master station received. 2) Integrate the system frequency, voltage level, the amount of generator tripping at each master station, and the system operation conditions to carry out system security checking. When problems such as excessive total generator tripping amount and wide range occur in a short time, resulting in exceeding the system frequency and voltage support regulation capabilities, a locking instruction is sent to the relevant regional master stations to effectively prevent and control broadband oscillation on the basis of ensuring the safety of the large power grid.

[0222] Regional Master Station: 1) Since the wide-frequency oscillations in the new power system are mainly caused by the interaction between power electronic devices such as wind power, photovoltaic power, and HVDC and the power grid, and the propagation range of oscillation events is regional, regional master stations are set up in areas with concentrated access to relevant new energy sources, such as important 750 kV collection substations in areas with concentrated access to new energy sources such as Hami, Dunhuang, and Xiazhou, or important node 220 kV new energy collection substations. 2) Organize the research and development of wide-frequency oscillation monitoring and source tracing technology, and based on the key oscillation characteristic quantities sent by each substation, locate the oscillation source within the region and implement control measures in the shortest possible time. 3) The regional master station has an online oscillation monitoring and analysis function that can receive various oscillation characteristics sent by each substation and locate the oscillation source. It can identify the oscillation source information within 1 second after the oscillation occurs through comprehensive comparison of oscillation characteristics, select the generator tripping according to the oscillation source location, and send the generator tripping command to the new energy substation;

[0223] Before sending the generator tripping command, send the generator tripping quantity to the dispatching main station for safety check. If no blocking command is received from the dispatching main station after a certain period of time, directly send it to the substation for generator tripping, providing a means for the timely and effective disposal of wide-frequency oscillation problems.

[0224] New Energy Substation: 1) Deployed at new energy power stations, equipped with an online oscillation monitoring and analysis device with high-precision sampling and millisecond-level computing capabilities (upgrading using a stability control or wide-frequency monitoring device, or a new device can also be installed). 2) It can conduct on-site monitoring and analysis of wide-frequency oscillations. By real-time monitoring and analyzing data such as voltage, current, and oscillation frequency of power generation units or feeders in wind farms or photovoltaic power plants, it can calculate oscillation characteristic quantities such as oscillation amplitude, oscillation start time, frequency spectrum, and transient energy in real time, and send relevant signals to the master station when the action threshold value is reached. 3) Receive and execute the generator tripping command sent by the regional master station.

[0225] In summary, by means of the above technical solutions of the present invention, the present invention monitors and calculates data such as voltage, current, and oscillation frequency of power generation units or feeders in a wind farm or a photovoltaic power station, calculates oscillation characteristic quantities such as oscillation amplitude, oscillation starting time, frequency spectrum, and transient energy in real time, and adds indicators such as voltage active and reactive power sensitivities for the first time to improve the accuracy of oscillation source analysis and positioning. An online oscillation monitoring and analysis function with the ability to receive various oscillation characteristics sent by each sub-station and perform oscillation source positioning is available in the main station in the construction area of the new energy centralized access and collection hub station. It can accurately identify the oscillation source and key participating power stations within seconds after the oscillation occurs through comprehensive comparison of oscillation characteristics. According to the oscillation source positioning, generator tripping selection is carried out and generator tripping commands are sent to the new energy sub-stations. In the main station constructed in the dispatching center, system security checks are carried out in combination with system frequency, voltage level, the generator tripping amount of each main station, and the system operation conditions. When problems such as excessive total generator tripping amount and over-wide range occur in a short time, exceeding the system frequency and voltage support and regulation capabilities, a blocking command is sent to the relevant regional main station to achieve effective prevention and control of broadband oscillation on the basis of ensuring the safety of the large power grid. In order to effectively cope with the impact of broadband oscillation on the power system, the present invention further designs a coordinated defense mechanism. This mechanism not only relies on the monitoring and prevention and control of individual stations, but also includes the coordinated cooperation between various new energy power stations. By establishing a regional main station to receive and integrate real-time monitoring data from each sub-station, the system can quickly locate the oscillation source and reasonably dispatch various devices in the power system according to the characteristics and influence range of the oscillation source, and take coordinated defense measures.

[0226] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, optical storage, etc.) containing computer-usable program code.

[0227] The above specific embodiments have further elaborated on the purpose, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A broadband oscillation on-site monitoring and coordinated defense system, characterized in that: include: New energy substation, used to obtain the electrical parameters of the new energy station and calculate the oscillation characteristic quantity of the new energy station according to the electrical parameters; When the oscillation characteristic quantity reaches the preset oscillation threshold, the electrical parameters and the oscillation characteristic quantity are sent to the regional master station, and the power-off instruction issued by the regional master station is received and executed; The regional master station is used to receive the electrical parameters and oscillation characteristics sent by the new energy substation and locate the oscillation source using broadband oscillation monitoring and tracing technology; Select the machine to be switched based on the oscillation source location result, and issue a switch command to the new energy substation; The dispatching master station is used to monitor the power system frequency, voltage level and the information on the amount of machine cutting of the master station in each region in real time, and to perform safety verification based on the power system frequency, voltage level and the information on the amount of machine cutting of the master station in each region, and to carry out power coordination defense based on the verification results.

2. A broadband oscillation on-site monitoring and coordinated defense system according to claim 1, characterized in that: The new energy substation includes: A data acquisition and analysis unit is used to obtain electrical parameters of the new energy station and identify oscillation events and oscillation event patterns in the new energy station by analyzing the development trend of the electrical parameters; The regional master station sending unit is used to send the electrical parameters of the new energy station to the regional master station after oscillation occurs in the new energy station; The receiving and executing unit is used to receive and execute the machine switching instruction sent by the regional master station.

3. A broadband oscillation on-site monitoring and coordinated defense system according to claim 2, characterized in that: The data acquisition and analysis unit comprises: A data monitoring unit, which is used to monitor and analyze the electrical parameters of the wind farm, photovoltaic power station generating units and feeders in real time using an oscillation online monitoring and analysis device based on a time division multiplexing method and an adaptive sampling frequency technology. The electrical parameters include voltage, current, active power, reactive power and apparent power. The window characteristic factor extraction unit is used to analyze the voltage and current angle and power curve fluctuation characteristics according to the electrical parameters of the wind farm, photovoltaic power station power generation unit and feeder, and extract the window characteristic factor by using data statistics and deduction methods; The oscillation event recognition unit is used to identify oscillation events based on window characteristic factors, through data statistics and trend analysis, and the combination of multi-factor weight ratio fusion luminescence method; The oscillation event pattern recognition unit is used to perform data enhancement processing on electrical parameters through a data-enhanced multimodal fusion strategy, and to build an oscillation pattern recognition model by integrating transfer learning and generative adversarial networks, and to use the oscillation pattern recognition model to identify oscillation event patterns.

4. A broadband oscillation on-site monitoring and coordinated defense system according to claim 3, characterized in that: According to the electrical parameters of the wind farm, photovoltaic power station power generation unit and feeder, the voltage and current angle and power curve fluctuation characteristics are analyzed, and the window characteristic factors are extracted by using data statistics and deduction methods to form the oscillation risk assessment elements, including: Determine the phase angle difference between the voltage phase angle and the current phase angle based on the electrical parameters of the wind farm, photovoltaic power station power generation unit and feeder, and analyze the changing trend of the phase angle difference; Perform fluctuation analysis on the curves of active power, reactive power and apparent power in the electrical parameters of wind farms, photovoltaic power station generating units and feeders, and identify the fluctuation characteristics of power curves; Based on the preset sliding window length, variable window length disturbance analysis is performed on the electrical parameters of wind farms, photovoltaic power station power generation units and feeders to identify the fluctuation characteristics of power disturbances within the window period; According to the changing trend of phase angle difference, the curve fluctuation characteristics of power and the fluctuation characteristics of power disturbance, the window characteristic factors are extracted by using data statistical methods, and the extracted window characteristic factors are used as the oscillation risk assessment factors.

5. A broadband oscillation on-site monitoring and coordinated defense system according to claim 3, characterized in that: The oscillation event identification based on the window characteristic factor, through data statistics and trend analysis, and the combination of multi-factor weight ratio fusion luminescence method includes: According to the extracted window characteristic factors, the elements of oscillation risk assessment are determined, and the corresponding weights are assigned according to the importance of each element to the oscillation risk assessment; The multi-factor weighted ratio fusion luminescence method is used to fuse the results of each factor and present the fusion structure in numerical form to form a quantitative risk value; It is determined whether the quantitative risk value exceeds the preset risk threshold. If so, it indicates that an oscillation event has occurred in the new energy station. Otherwise, it indicates that no oscillation event has occurred in the new energy station.

6. A broadband oscillation on-site monitoring and coordinated defense system according to claim 3, characterized in that: The multi-modal fusion strategy of data enhancement is used to perform data enhancement processing on the electrical parameters, and the transfer learning and the generative adversarial network are integrated to construct an oscillation pattern recognition model. The oscillation pattern recognition model is used to identify the oscillation event mode, including: The data transformation method is used to transform the electrical parameter data to generate preliminary data enhancement samples; Based on the preliminary data enhancement samples, new electrical parameter data samples are generated through this simulated disturbance based on circuit laws such as Kirchhoff's laws and electromagnetic principles; Collect external environmental data related to new energy stations, and fuse new electrical parameter data samples with external environmental data to form a multimodal data set; Using transfer learning technology, the pre-trained convolutional neural network model is transferred to the oscillation pattern recognition task; Build a generative adversarial network model to generate high-quality simulated oscillation data samples, expand the training data set, use the multimodal data set and the expanded training data set to train the oscillation pattern recognition model, and use the oscillation pattern recognition model to identify oscillation event patterns.

7. A broadband oscillation on-site monitoring and coordinated defense system according to claim 1, characterized in that: The regional master station includes: An oscillation data receiving unit, used to receive electrical parameters and oscillation characteristic quantities of a new energy station sent by a new energy substation; An oscillation data processing unit, used to calculate an oscillation characteristic quantity according to the electrical parameters of the new energy station sent by the new energy substation, and determine an oscillation source discrimination condition according to the oscillation characteristic quantity; A vibration source determination unit, used to determine the vibration source in the new energy station according to the vibration source auxiliary discrimination conditions; The oscillation source action output giving unit is used to determine whether the oscillation source diverges and whether the oscillation duration is greater than a preset time threshold, and to sort the comprehensive characteristic values ​​of the new energy stations participating in the oscillation, send a power-off command signal to the new energy substation, and calm the oscillation according to the comprehensive characteristic value sorting result; The machine cutting quantity uploading unit is used to send the machine cutting quantity information to the dispatching main station at the same time when sending the machine cutting instruction signal to the new energy substation.

8. A broadband oscillation on-site monitoring and coordinated defense system according to claim 7, characterized in that: The calculating of the oscillation characteristic quantity according to the electrical parameters of the new energy station sent by the new energy substation, and determining the oscillation source identification condition according to the oscillation characteristic quantity includes: Based on the adaptive normalization algorithm of data entropy, the electrical parameters of the new energy station are normalized to obtain a normalized data set; according to the normalized data set, the amplitude and start-up time of each device in the new energy station are determined through the time analysis window, and the amplitude and start-up time of each device in the new energy station are sorted to obtain the sorting of the start-up time of each device; According to the oscillation mode frequency in the normalized data set, the oscillation mode frequency is clustered, and the positive sequence impedance of each oscillation mode is calculated according to the clustering result to obtain the voltage phasor and current phasor under the oscillation mode frequency; according to the positive sequence impedance characteristic calculation formula, the positive sequence impedance characteristic of the new energy station equipment under the oscillation mode is calculated; The positive sequence impedance characteristic calculation formula is: In the formula, represents the positive sequence impedance under each oscillation mode, represents the voltage phasor, represents the current phasor, j represents the imaginary part, R m represents the resistance in each oscillation mode, X m represents the reactance in each oscillation mode; Comparing the apparent power in the normalized data set with a preset apparent power threshold, and comparing the oscillation amplitude in the normalized data set with a preset oscillation amplitude threshold, to obtain a comparison result; The oscillation start-up time sequencing of each device, the positive-sequence impedance characteristics of the new energy station equipment in the oscillation mode and the comparison results are used as auxiliary judgment conditions for the oscillation source.

9. A broadband oscillation on-site monitoring and coordinated defense system according to claim 8, characterized in that: The adaptive normalization algorithm based on data entropy normalizes the electrical parameters of the new energy station to obtain a normalized data set including: For each characteristic quantity in the electrical parameters of the new energy station, calculate the characteristic quantity information entropy; The calculation formula of the feature quantity information entropy is: In the formula, H(x) represents the information entropy of feature x, p(x i ) represents the probability of the occurrence of the i-th feature x, and n represents the number of feature types; Determine a scaling factor according to the characteristic quantity information entropy, and use the scaling factor to normalize each electrical parameter to obtain a normalized data set; The calculation formula of the scaling factor is: Where, X norm represents the scaling factor, x represents the characteristic quantity in the electrical parameter, min(x) represents the minimum value of the characteristic quantity in the electrical parameter, max(x) represents the maximum value of the characteristic quantity in the electrical parameter, H(x) represents the information entropy of the characteristic quantity x, and e represents a constant.

10. A broadband oscillation on-site monitoring and coordinated defense system according to claim 8, characterized in that: The clustering process of the oscillation mode frequencies comprises: The oscillation modal frequency data are standardized using the Z-score method to obtain standardized oscillation modal frequency data; According to the standardized oscillation mode frequency data, the Euclidean distance between each oscillation mode frequency data is calculated using the Euclidean distance; The Euclidean distance between each oscillation mode frequency data is used as the input of the clustering algorithm, and the Euclidean distance between each oscillation mode frequency data is clustered by the density peak clustering algorithm to obtain the clustering result; The clustering process of the Euclidean distances between the oscillation mode frequency data by using the density peak clustering algorithm includes: constructing a distance matrix according to the Euclidean distances between the oscillation mode frequency data; For each oscillation mode frequency data point in the distance matrix, its local density is calculated based on the Gaussian kernel function method; For each oscillation mode frequency data point, calculate the minimum distance between it and all points with higher density than it; The cluster center is selected according to the local density and the minimum distance, and other points are assigned to the corresponding clusters according to the density propagation principle to obtain the final clustering result.

Images