Relay protection simulation method and system

Abstract

The invention discloses a relay protection simulation method and system, and the method comprises the steps: carrying out the modular design of a relay protection simulation system, and controlling the starting or disabling of each protection module in the system through a configuration file; the method comprises the following steps: acquiring multi-source heterogeneous data from an actual power grid and a simulation test bed, preprocessing, establishing a model, and learning and classifying fault features of equipment; enhancing and optimizing the model, performing a relay protection simulation test, and controlling a simulation system to simulate a fault scene, including adjusting the position of a fault point, changing a fault type, setting different fault parameters, simulating an abnormal working condition, recording simulation data in real time, and starting a protection module to simulate relay protection; simulation relay protection is optimized through an analytic calculation method, and the cooperation relation between the relay protection is analyzed; modularized relay protection is applied to an actual power grid. According to the relay protection simulation method provided by the invention, a simulation test is carried out in combination with the model, and the cooperation relationship between different relay protections is evaluated.

Description

Technical Field

[0001] The present invention relates to the field of electrical digital data processing technology, and in particular to the field of using machine learning training models, and specifically to a relay protection simulation method and system. Background Art

[0002] Relay protection is an automated measure and equipment used to protect the power system and its main equipment, or to reflect the failure or abnormal working conditions of the power system equipment, and directly act on the circuit breaker to trip to terminate the development of these events, or send a timely signal to the on-duty operating personnel when the power system components or the power system itself fail or an event threatens its safe operation. When a power system failure or abnormality occurs, the faulty equipment is automatically removed from the system in the shortest time and smallest area, or a signal is sent to the on-duty personnel to eliminate the root cause of the abnormal condition, so as to avoid or reduce damage to the equipment and reduce the impact on power supply to adjacent areas. Relay protection mainly uses the changes in electrical quantities when short circuits or abnormal conditions occur in components in the power system to constitute the physical quantities of relay protection action.

[0003] At present, the development of new energy equipment is very rapid. Since it uses renewable energy or new energy for energy conversion, storage and utilization, it reduces dependence on traditional fossil fuels, reduces environmental pollution and improves energy utilization efficiency. More and more new energy equipment is connected to the power grid for operation. Compared with traditional equipment, there is a big difference between the electrical quantity of new energy equipment and traditional equipment. Therefore, there needs to be a difference in the relay protection for new energy equipment. The working characteristics of traditional synchronous generators are basically linear, that is, the output voltage and current are linearly related with the load changes, which is easy to control and predict. However, new energy equipment has nonlinear characteristics. This nonlinear characteristic will cause the traditional relay protection algorithm to fail and affect the accuracy and speed of fault detection.

[0004] Therefore, it is necessary to improve the relay protection simulation method and system in the prior art to solve the above problems. Summary of the invention

[0005] The present invention overcomes the deficiencies of the prior art and provides a relay protection simulation method and system, aiming to solve the problems of decreased fault detection accuracy and response speed faced by the prior art when new energy equipment is connected to the power grid.

[0006] To achieve the above object, the technical solution adopted by the present invention is: a relay protection simulation method, comprising:

[0007] S1. Modular design of relay protection simulation system, controlling the enabling or disabling of each protection module in the system through configuration files;

[0008] S2, obtain multi-source heterogeneous data from the actual power grid and simulation test bench, perform preprocessing, build models, and learn and classify the fault characteristics of equipment;

[0009] S3. Enhance and optimize the model, conduct relay protection simulation test, control the simulation system to simulate fault scenarios, including adjusting the location of the fault point, changing the fault type, setting different fault parameters and simulating abnormal working conditions, record simulation data in real time, and start the protection module to simulate relay protection;

[0010] S4. Optimize the simulated relay protection by analytical calculation method and analyze the coordination relationship between the relay protections;

[0011] S5. Apply modular relay protection to actual power grids.

[0012] In a preferred embodiment of the present invention, in step S1, the design step of the simulation system includes:

[0013] S11. Determine the type of relay protection function according to the characteristics of the power grid structure, the characteristics of new energy equipment and operation requirements, and divide it into independent protection modules;

[0014] Types of relay protection functions include: longitudinal current differential protection, distance protection and zero sequence protection;

[0015] S12. Define the interface standards and communication protocols of each protection module to ensure smooth information exchange;

[0016] S13. Control the enabling, disabling and setting adjustment of the protection module by setting a configuration file.

[0017] In a preferred embodiment of the present invention, the simulation system adopts primary and secondary model joint simulation, and a custom data interaction port is designed in PSCAD to realize data transmission between the primary system and the secondary system;

[0018] The primary system includes: power supply, transmission lines, transformers, loads, circuit breakers and fault points; the secondary system includes: relay protection devices, voltage transformers, current transformers, data acquisition and processing systems, control equipment and communication equipment; the power supply is new energy stations and traditional synchronous generators.

[0019] In a preferred embodiment of the present invention, in step S2, the pre-processing step includes:

[0020] S21, remove outliers caused by sensor failure and communication interference;

[0021] S22, using sliding window technology to capture transient information at the moment of fault occurrence;

[0022] S23, classifying the data samples into normal and fault labels according to the fault status;

[0023] S24. Enhance and standardize data samples.

[0024] In a preferred embodiment of the present invention, in step S2, the overall architecture of the model includes: an input layer, a CNN feature extraction layer, an LSTM time series analysis layer, a fully connected layer and an output layer; the steps of establishing the model include:

[0025] S25, using the pre-processed multi-source heterogeneous data as input data of the input layer, including: current, voltage, frequency and power;

[0026] S26, use a 1D convolutional layer to extract features from the input data, and use the ReLU activation function for nonlinear mapping;

[0027] S27, downsample the output of the convolutional layer through the maximum pooling layer to reduce the number of parameters and avoid overfitting;

[0028] S28, using the output result of the convolutional neural network as the input data of the long short-term memory network, the internal calculation of the unit of the long short-term memory network includes a forget gate, an input gate, a cell state update and an output gate;

[0029] S29. After flattening the output of the long short-term memory network, nonlinear mapping is performed through the fully connected layer, and the softmax activation function is applied to the output of the fully connected layer to obtain the probability distribution of each fault category.

[0030] In a preferred embodiment of the present invention, in step S3, the network weights are updated using the gradient descent method to perform model training optimization. Among them, θ represents the model parameters, J(θ) is the loss function, η is the learning rate, is the gradient of the parameter, and t is the number of current iterations.

[0031] In a preferred embodiment of the present invention, in step S3, the step of relay protection simulation test includes:

[0032] S31. Set the fault point in the relay protection simulation system and configure the fault type; the fault point can be set at different positions and in different quantities to test the coordination relationship between different protection devices; the fault types include: phase-to-phase short circuit, ground fault, etc.;

[0033] S32, running the simulation model, observing the operation of each relay protection device, and recording the state change of the circuit breaker and the alarm information of the relay protection device;

[0034] S33. Simulate abnormal operating conditions by controlling the equipment in the secondary system model.

[0035] In a preferred embodiment of the present invention, the abnormal operating conditions include: abnormality of the relay protection device, communication interruption, mutual inductor disconnection, circuit breaker failure and DC voltage loss.

[0036] In a preferred embodiment of the present invention, in step S4, the coordination relationship between the relay protections includes: full coordination, conditional coordination, incomplete coordination and complete non-coordination;

[0037] Evaluate the coordination between various relay protection devices based on the data recorded by simulation;

[0038] S41. Check whether there is any situation where the relay protection device fails to operate as expected;

[0039] S42, calculating the action time difference between the protection devices, and determining whether it complies with the preset coordination logic;

[0040] S43, calculating the efficiency and accuracy of fault removal;

[0041] S44. Based on the data results, evaluate the coordination effect between the relay protections and determine the type of coordination relationship;

[0042] In step S44, the coordination effect between the relay protections is evaluated through risk assessment;

[0043] Key indicators reflecting the coordination effect of relay protection include: action time difference, fault removal time, false operation rate and missed detection rate;

[0044] Action time difference risk indicator: ΔT risk =max(ΔT ij )-Δt th , where ΔT ij is the operating time difference between relay protection devices i and j, Δt th is the acceptable threshold of action time difference;

[0045] Fault removal time risk index: T cl =T cla -T cli , where T cla is the actual fault clearing time, T cli is the ideal fault removal time;

[0046] Misoperation rate risk indicators: Among them, N f is the number of false trips, N t is the total number of actions;

[0047] Missed detection rate risk indicator:

[0048] The above risk indicators are weighted and summed to obtain a comprehensive risk indicator: Among them, ω i is the weight of the ith risk indicator, R i is the value of the ith risk indicator, and n is the number of risk indicators.

[0049] The present invention provides a relay protection simulation system, comprising:

[0050] Modular protection function modules, including: longitudinal current differential protection module, distance protection module, zero sequence protection module and other protection modules, are used to simulate the protection methods in the actual power system and perform fault detection and removal on the power system;

[0051] Data preprocessing module, processing the collected raw data;

[0052] The deep learning model module uses CNN and LSTM to extract features and analyze time series of preprocessed data to identify and classify faults;

[0053] The simulation test module sets fault points and abnormal conditions in the simulation environment, runs the model and observes the action of the protection device;

[0054] The coordination relationship analysis module analyzes the coordination relationship between protection devices, identifies weak links through risk assessment, and optimizes and adjusts the relay protection system.

[0055] The present invention solves the defects existing in the background technology and has the following beneficial effects:

[0056] (1) The present invention proposes a relay protection simulation method and system, which improves the adaptability and test and evaluation capabilities of new energy equipment connected to the power grid by modularly designing the relay protection system and combining deep learning technology. First, a unified platform containing independent protection modules such as longitudinal current differential protection, distance protection and zero-sequence protection is constructed, and the enabling or disabling of each module is flexibly controlled through the configuration file; then, multi-source heterogeneous data is obtained from the actual power grid and the simulation test bench, and after preprocessing, it is used to train a model that combines convolutional neural networks and long short-term memory networks to identify and classify fault characteristics; then, the gradient descent method is used to optimize the model parameters, and simulation tests are carried out. At the same time, analytical calculation methods are used to evaluate the coordination relationship between different relay protections and their performance indicators such as accuracy, response speed and missed detection rate; finally, the protection strategy is adjusted based on the simulation results, and the optimized relay protection scheme is applied to the actual power grid to ensure the stable operation of the power system.

[0057] (2) The present invention combines a modular design of a relay protection simulation system with a model established by combining a convolutional neural network and a long short-term memory network. Different protection functions are operated as independent modules in a unified platform, and a deep learning model is used to learn and classify the fault characteristics of the equipment. This not only improves the accuracy of fault detection, but also enables the system to flexibly adapt to the diversity and complexity of new energy equipment. Compared with the prior art, it further achieves the effect of improving the adaptability of the relay protection system to the access of new energy equipment to the power grid and the test and evaluation capabilities.

[0058] (3) The present invention combines the model established by combining convolutional neural networks and long short-term memory networks with analytical calculation methods, the advantages of deep learning models in extracting fault characteristics, and how to quantitatively evaluate simulation results through analytical calculation methods. It not only improves the accuracy and response speed of fault detection, but also provides a deep understanding and optimization direction of the performance of the relay protection system. Compared with the existing technology, it further achieves the effect of enhancing the performance optimization and fault detection capability of the relay protection system.

[0059] (4) The present invention combines multi-source heterogeneous data preprocessing with model building based on convolutional neural networks and long short-term memory networks, uses interquartile range to identify and remove outliers, adopts sliding window technology to capture transient information, and uses CNN-LSTM model to learn the fault characteristics of the power system. This method not only optimizes the input data quality, but also enhances the model's ability to learn complex fault modes, so that the system can provide more accurate fault classification results when facing nonlinear and intermittent problems unique to new energy equipment. Compared with the existing technology, it further improves the fault diagnosis capability under complex working conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art are briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work.

[0061] Figure 1 is a flow chart of a preferred embodiment of the present invention;

[0062] Figure 2 It is a schematic diagram of a method of a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0063] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0064] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0065] Application Overview:

[0066] The existing relay protection simulation test methods can usually only test a single relay protection principle, and cannot fully evaluate the coordination of multiple protection principles under complex working conditions. The nonlinear characteristics and limited short-circuit current characteristics of new energy equipment cause the failure of traditional relay protection algorithms, affecting the accuracy and speed of fault detection. In addition, the intermittent and uncertain nature of new energy equipment increases the complexity of system operation, which may cause fluctuations in system frequency and voltage, affecting the transient stability of the system. Therefore, the existing relay protection simulation methods have obvious deficiencies in adapting to the test and evaluation of new energy equipment connected to the power grid.

[0067] This application proposes a relay protection simulation method, which aims to address the limitations of single relay protection principle testing in the prior art and improve the adaptability testing and evaluation capabilities for new energy equipment connected to the power grid.

[0068] Exemplary Methods

[0069] like Figure 1 and Figure 2 As shown, a relay protection simulation method comprises the steps of:

[0070] S1. Modular design of relay protection simulation system, controlling the enabling or disabling of each protection module in the system through configuration files;

[0071] S2, obtain multi-source heterogeneous data from the actual power grid and simulation test bench, perform preprocessing, build models, and learn and classify the fault characteristics of equipment;

[0072] S3. Enhance and optimize the model, conduct relay protection simulation test, control the simulation system to simulate fault scenarios, including adjusting the location of the fault point, changing the fault type, setting different fault parameters and simulating abnormal working conditions, record simulation data in real time, and start the protection module to simulate relay protection;

[0073] S4. Optimize the simulated relay protection by analytical calculation method and analyze the coordination relationship between the relay protections;

[0074] S5. Apply modular relay protection to actual power grids.

[0075] The modular relay protection simulation system integrates different protection functions into a unified platform or framework, and each protection function runs as an independent module. Since new energy equipment has different nonlinear characteristics and short-circuit current characteristics, the modular relay protection system can provide special protection strategies for different types of equipment.

[0076] In step S1, the design steps of the simulation system include:

[0077] S11. According to the characteristics of the power grid structure, the characteristics of new energy equipment and the operation requirements, determine the type of relay protection function and divide it into independent protection modules; the types of relay protection functions include: longitudinal current differential protection, distance protection and zero-sequence protection;

[0078] S12. Define the interface standards and communication protocols of each protection module to ensure smooth information exchange;

[0079] S13, controlling the enabling, disabling and protection value of the protection module by setting a configuration file;

[0080] The system of this application adopts joint simulation of primary and secondary models, and designs a custom data interaction port in PSCAD to realize data transmission between the primary system and the secondary system; the primary system is responsible for the generation, transmission, distribution and use of electric energy, and is the infrastructure of the power system. The main function of the secondary system is to ensure the safe, stable and efficient operation of the primary system, and to respond to various faults and abnormal situations through real-time monitoring, control and protection; through the data interface, the simulation data calculated in real time by the primary system is transmitted to the secondary system, and the calculation results of the secondary system are returned to the primary system, forming a closed-loop system to truly simulate the action process of the protection device.

[0081] Specifically, the primary system model includes: power supply, transmission line, transformer, load, circuit breaker and fault point; the secondary system includes: relay protection device, voltage transformer, current transformer, data acquisition and processing system, control equipment and communication equipment; the power supply is new energy station and traditional synchronous generator power supply;

[0082] Longitudinal current differential protection is a relay protection method based on the current differential principle. When new energy equipment is connected to the power grid, its nonlinear characteristics and intermittent working mode will cause current waveform distortion or instability. Longitudinal current differential protection can accurately identify abnormal current changes caused by new energy equipment by comparing the current difference at both ends of the line, thereby eliminating the fault in time.

[0083] Distance protection is a relay protection method that determines the fault and cuts off the faulty part based on the distance from the fault point to the protection installation. After the new energy equipment is connected to the power grid, the traditional distance protection algorithm will fail due to its limited short-circuit current characteristics. By improving the distance protection algorithm and considering the short-circuit current characteristics of the new energy equipment, it can calculate the distance to the fault point more accurately.

[0084] Zero-sequence protection is a relay protection method specifically used to detect and handle ground faults in power systems. When new energy equipment is connected to the power grid, ground faults will frequently occur due to its special grounding method or fault mode. Zero-sequence protection can detect and handle these ground faults in a timely manner by monitoring changes in zero-sequence current, thereby protecting the safety of the system and equipment.

[0085] Each protection module is independent and can be maintained, upgraded or replaced individually without affecting the normal operation of other modules, reducing maintenance costs and complexity. In addition, with the increase in new energy equipment and the complexity of the grid structure, the modular design allows the system to be easily expanded to adapt to new protection needs and technological developments.

[0086] Multi-source heterogeneous data refers to data from different sources with different formats, structures and characteristics. By integrating multi-source data, the actual operating conditions of the power system can be more comprehensively reflected and the accuracy and credibility of the simulation can be improved.

[0087] In step S2, the data types include: current, voltage, frequency and power;

[0088] The actual grid operation data includes real-time monitored current, voltage and frequency data, which have time series characteristics; the simulation test bench data refers to the data obtained by simulation software simulating power system faults and abnormal conditions, which can be used to verify and protect the performance of the algorithm.

[0089] The data of the actual power grid can also be simulated through actual equipment for detection; specifically, the SD2820B distribution network primary and secondary fusion equipment test device provides high-precision three-phase voltage up to 19kV and high current up to 1000A, and has the ability to simulate primary current and primary voltage circuits; the three-phase 1000A high current test can be used to simulate high current conditions in the power grid, and the three-phase 10kV high voltage test can be used to simulate the behavior of the power system under high voltage environment. Combined with the P2200A5 distribution automation terminal tester, the working status of the distribution automation terminal equipment can be checked, and it can be used as an auxiliary tool to record relevant power grid data.

[0090] In step S2, the preprocessing steps include:

[0091] S21, remove outliers caused by sensor failure and communication interference;

[0092] S22, using sliding window technology to capture transient information at the moment of fault occurrence;

[0093] S23, classifying the data samples into normal and fault labels according to the fault status;

[0094] S24, enhance and standardize data samples;

[0095] In step S21, the interquartile range is used to identify and remove outliers.

[0096] In step S22, the window size is set to W, and the window is slid along the time series to extract the features in the window. The window starting from time point t can be expressed as: X t:W =[x t ,x t+1 ,…,x t+W-1 ] and then perform statistical analysis on the data in each window.

[0097] In step S24, white noise is added to the data for data enhancement; the data is Z-score standardized. Among them, x i is the original data, μ is the mean of all data, and σ is the standard deviation of all data.

[0098] Step S2 ensures that the multi-source heterogeneous data obtained from the actual power grid and simulation test bench can be effectively used for relay protection simulation testing. Through preprocessing, the quality of the data is improved, outliers and noise are removed or weakened, the timing characteristics and fault characteristics of the data are better captured, and the data are unified to the same scale to facilitate subsequent model training and simulation testing.

[0099] In step S3, since new energy equipment has nonlinear characteristics and intermittent working mode, its fault characteristics change over time. The convolutional neural network and long short-term memory network are combined to build a model. CNN has advantages in extracting spatial characteristics of data and can capture the local patterns of current and voltage waveforms; while LSTM performs well in extracting temporal characteristics of data and can capture the changing trend of current and voltage over time. By combining the two, the spatial and temporal information of the data can be fully utilized to improve the model's ability to learn and classify the fault characteristics of new energy equipment.

[0100] A deep learning model combining convolutional neural network and long short-term memory network is used for end-to-end fault feature learning and diagnosis classification. The overall architecture of the model includes: input layer, CNN feature extraction layer, LSTM time series analysis layer, fully connected layer and output layer;

[0101] In step S2, the steps of establishing the model include:

[0102] S25. The preprocessed multi-source heterogeneous data is used as the input data of the input layer, including current, voltage, frequency and power, represented as X∈R N×T×C , where N is the number of samples, T is the length of the time series, and C is the dimension of the feature;

[0103] S26, use a 1D convolutional layer to extract features from the input data, with a convolution kernel size of k, a step size of s, and a padding method of valid; convolution operation Y conv =Conv1D(X,k,s), where Conv1D is a 1D convolution operation, using the ReLU activation function for nonlinear mapping, Y relu =ReLU(Y conv ); the activation output value is Among them, W is the convolution kernel weight, b is the bias term, and i,j is the i row and j column on the neural network output feature map.

[0104] S27, downsample the output of the convolutional layer through the maximum pooling layer, the pooling window size is p, the step size is p, and the output value of the pooling operation is Y pool (i,j)=max(Y relu (i,j:j+p-1)).

[0105] S28. Use the output of the CNN part as the input of the LSTM, and set the number of LSTM units to h; LSTM operation Y lstm =LSTM(Y pool ,h), the internal calculation of the LSTM unit includes the forget gate, input gate, cell state update and output gate; the final output result is the hidden state update h t =o t ×tanh(C t ), where o t is the output gate result, C t Update the state for the cell.

[0106] S29, after flattening the output of LSTM, nonlinear mapping is performed through the fully connected layer, and the softmax activation function is applied to the output of the fully connected layer to obtain the probability distribution of each fault category;

[0107] New energy equipment, such as wind power generation and solar power generation, has nonlinear characteristics and intermittent working modes. Its fault characteristics are not only complex but also change over time. Traditional methods are difficult to effectively capture these characteristics. Convolutional neural networks are good at extracting spatial characteristics of data and can capture the local patterns of current and voltage waveforms of new energy equipment. Long short-term memory networks can capture the changing trend of data over time and the changes of current and voltage parameters over time, which is convenient for understanding the dynamic behavior of equipment and predicting faults. By combining CNN and LSTM to build a model, we can make full use of the advantages of both and improve the learning and classification capabilities of new energy equipment fault characteristics.

[0108] In step S3, the gradient descent method is used to update the network weights for model training optimization. Among them, θ represents the model parameters, J(θ) is the loss function, η is the learning rate, is the gradient of the parameter, and t is the number of current iterations.

[0109] In step S3, the steps of relay protection simulation test include:

[0110] S31. Set the fault point in the relay protection simulation system and configure the fault type; the fault point can be set at different positions and in different quantities to test the coordination relationship between different protection devices; the fault types include: phase-to-phase short circuit, ground fault, etc.;

[0111] S32, running the simulation model, observing the operation of each relay protection device, and recording the state change of the circuit breaker and the alarm information of the relay protection device;

[0112] S33, simulating abnormal operating conditions by controlling equipment in the secondary system model;

[0113] Abnormal working conditions include: abnormal relay protection device, communication interruption, transformer disconnection, circuit breaker failure and DC voltage loss;

[0114] Relay protection device abnormality can be simulated by modifying the control field in the configuration file to simulate partial or complete functional failure of the relay protection device; communication interruption can be simulated by setting the communication port between the relay protection devices to zero; mutual inductor disconnection can be simulated by setting the voltage or current signal transmitted to the relay protection device to zero; circuit breaker failure can be simulated by stopping the control signal output by the relay protection device to the circuit breaker and outputting a closing signal to the circuit breaker; DC voltage loss can be simulated by setting the DC voltage loss control port of the relay protection device to "1" to simulate battery damage;

[0115] According to different fault conditions, the protection module is started to simulate the relay protection, and the time, impact range and efficiency data of the simulated relay protection are recorded, and evaluation and ranking are carried out. The evaluation method adopts weighted evaluation.

[0116] Through the above steps, the performance of the relay protection system can be comprehensively evaluated to ensure that it can still work effectively when facing new challenges brought by the access of new energy equipment. In addition, these tests help identify potential problems and provide a basis for future improvements.

[0117] In the simulation of power system relay protection, it is not enough to rely on model training and simulation testing. The actual power system is very complex and involves the coordination relationship between multiple protection principles and equipment.

[0118] In step S4, the results of the simulated relay protection are analyzed using an analytical calculation method to analyze the accuracy, speed and missed detection rate of the model for typical faults;

[0119] Detection accuracy Among them, TP means that the fault is correctly detected, and FN means that the actual fault is not detected by false negative;

[0120] Response speed RT = t a -t f , where t a is the protection action time, t f is the time when the fault occurred;

[0121]

[0122] The above calculation method can help to quantitatively evaluate the performance of the relay protection system and guide further optimization work.

[0123] In step S4, relay protection verification is completed through analytical calculation methods to optimize relay protection risks.

[0124] According to the verification results, adjust the parameters of CNN and LSTM networks such as the number of layers, number of neurons, and learning rate to improve the accuracy and response speed of the model.

[0125] Step S4, the coordination relationship between the relay protections includes: full coordination, conditional coordination, incomplete coordination and no coordination at all;

[0126] Full coordination means that two or more relay protection devices can operate correctly according to the preset logic and timing to jointly eliminate the fault when a power grid fault occurs; conditional coordination means that relay protection devices can cooperate under specific conditions, but cannot cooperate under other conditions; incomplete coordination means that there is a time difference or logical inconsistency in the action of relay protection devices, but they are not completely ineffective, resulting in prolonged fault elimination time or certain impact on power grid stability; complete lack of coordination means that relay protection devices cannot work together and cannot effectively eliminate the fault when a fault occurs.

[0127] Evaluate the coordination between various relay protection devices based on the data recorded by simulation;

[0128] S41. Check whether there is any situation where the relay protection device fails to operate as expected;

[0129] S42, calculating the action time difference between the protection devices, and determining whether it complies with the preset coordination logic;

[0130] S43, calculating the efficiency and accuracy of fault removal;

[0131] S44. Based on the above data results, evaluate the coordination effect between the relay protections and determine the type of coordination relationship;

[0132] In step S44, the coordination effect between the various relay protections is evaluated through risk assessment. Risk assessment can help identify the weak links of the relay protection system when facing different fault scenarios, and whether the coordination between the various protection devices meets the expected safety standards.

[0133] Select key indicators that can reflect the coordination effect of relay protection, including: action time difference, fault removal time, false operation rate and missed detection rate;

[0134] Action time difference risk indicator: ΔT risk =max(ΔT ij )-Δt th , where ΔT ij is the operating time difference between relay protection devices i and j, Δt th is the acceptable threshold of action time difference;

[0135] Fault removal time risk index: T cl =T cla -T cli , where T cla is the actual fault clearing time, T cli is the ideal fault removal time;

[0136] Misoperation rate risk indicators: Among them, N f is the number of false trips, N t is the total number of actions;

[0137] Missed detection rate risk indicator:

[0138] The above risk indicators are weighted and summed to obtain a comprehensive risk indicator, which is used to comprehensively evaluate the coordination effect between various relay protections; the risk calculation formula is: Among them, ω i is the weight of the ith risk indicator, R iis the value of the ith risk indicator, and n is the number of risk indicators.

[0139] According to the risk assessment results, the relay protection system is optimized and adjusted, including: adjusting the parameter settings of the protection device, optimizing the communication network and algorithm, and upgrading or replacing the protection device, so as to improve the coordination relationship, reduce communication delays and improve the accuracy of fault identification.

[0140] In step S5, based on the simulation test and optimization results, it is determined which relay protection modules need to be deployed or updated in the actual power grid.

[0141] Example systems:

[0142] A relay protection simulation system, comprising:

[0143] Modular protection function modules, including: longitudinal current differential protection module, distance protection module, zero sequence protection module and other protection modules, are used to simulate the protection methods in the actual power system and perform fault detection and removal on the power system;

[0144] Data preprocessing module, processing the collected raw data;

[0145] The deep learning model module uses CNN and LSTM to extract features and analyze time series of preprocessed data to identify and classify faults;

[0146] The simulation test module sets fault points and abnormal conditions in the simulation environment, runs the model and observes the action of the protection device;

[0147] The coordination relationship analysis module analyzes the coordination relationship between protection devices, identifies weak links through risk assessment, and optimizes and adjusts the relay protection system.

[0148] The above is based on the ideal embodiment of the present invention. Through the above description, relevant personnel can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the content in the specification, and the technical scope must be determined according to the scope of the claims.

Claims

1. A relay protection simulation method, characterized in that: Includes steps: S1. Modular design of relay protection simulation system, controlling the enabling or disabling of each protection module in the system through configuration files; S2, obtain multi-source heterogeneous data from the actual power grid and simulation test bench, perform preprocessing, build models, and learn and classify the fault characteristics of equipment; S3. Enhance and optimize the model, conduct relay protection simulation test, control the simulation system to simulate fault scenarios, including adjusting the location of the fault point, changing the fault type, setting different fault parameters and simulating abnormal working conditions, record simulation data in real time, and start the protection module to simulate relay protection; S4. Optimize the simulated relay protection by analytical calculation method and analyze the coordination relationship between the relay protections; S5. Apply modular relay protection to actual power grids.

2. A relay protection simulation method according to claim 1, characterized in that: In step S1, the design steps of the simulation system include: S11. Determine the type of relay protection function according to the characteristics of the power grid structure, the characteristics of new energy equipment and operation requirements, and divide it into independent protection modules; Types of relay protection functions include: longitudinal current differential protection, distance protection and zero sequence protection; S12. Define the interface standards and communication protocols of each protection module to ensure smooth information exchange; S13. Control the enabling, disabling and setting adjustment of the protection module by setting a configuration file.

3. A relay protection simulation method according to claim 2, characterized in that: The simulation system adopts joint simulation of primary and secondary models, and designs a custom data interaction port in PSCAD to realize data transmission between the primary system and the secondary system; The primary system includes: power supply, transmission lines, transformers, loads, circuit breakers and fault points; the secondary system includes: relay protection devices, voltage transformers, current transformers, data acquisition and processing systems, control equipment and communication equipment; the power supply is new energy stations and traditional synchronous generators.

4. A relay protection simulation method according to claim 1, characterized in that: In step S2, the preprocessing steps include: S21, remove outliers caused by sensor failure and communication interference; S22, using sliding window technology to capture transient information at the moment of fault occurrence; S23, classifying the data samples into normal and fault labels according to the fault status; S24. Enhance and standardize data samples.

5. A relay protection simulation method according to claim 1, characterized in that: In step S2, the overall architecture of the model includes: input layer, CNN feature extraction layer, LSTM time series analysis layer, fully connected layer and output layer; the steps of building the model include: S25, using the pre-processed multi-source heterogeneous data as input data of the input layer, including: current, voltage, frequency and power; S26, use a 1D convolutional layer to extract features from the input data, and use the ReLU activation function for nonlinear mapping; S27, downsample the output of the convolutional layer through the maximum pooling layer to reduce the number of parameters and avoid overfitting; S28, using the output result of the convolutional neural network as the input data of the long short-term memory network, the internal calculation of the unit of the long short-term memory network includes a forget gate, an input gate, a cell state update and an output gate; S29. After flattening the output of the long short-term memory network, nonlinear mapping is performed through the fully connected layer, and the softmax activation function is applied to the output of the fully connected layer to obtain the probability distribution of each fault category.

6. A relay protection simulation method according to claim 1, characterized in that: In step S3, the gradient descent method is used to update the network weights for model training optimization. Among them, θ represents the model parameters, J(θ) is the loss function, η is the learning rate, is the gradient of the parameter, and t is the number of current iterations.

7. A relay protection simulation method according to claim 3, characterized in that: In step S3, the steps of relay protection simulation test include: S31. Set the fault point in the relay protection simulation system and configure the fault type; the fault point can be set in different positions and different numbers to test the coordination relationship between different protection devices; the fault type includes: phase-to-phase short circuit, ground fault, etc.; S32, running the simulation model, observing the operation of each relay protection device, and recording the state change of the circuit breaker and the alarm information of the relay protection device; S33. Simulate abnormal operating conditions by controlling the equipment in the secondary system model.

8. A relay protection simulation method according to claim 7, characterized in that: Abnormal operating conditions include: abnormal relay protection device, communication interruption, transformer disconnection, circuit breaker failure and DC voltage loss.

9. A relay protection simulation method according to claim 1, characterized in that: Step S4, the coordination relationship between the relay protections includes: full coordination, conditional coordination, incomplete coordination and no coordination at all; Evaluate the coordination between various relay protection devices based on the data recorded by simulation; S41. Check whether there is any situation where the relay protection device fails to operate as expected; S42, calculating the action time difference between the protection devices, and determining whether it complies with the preset coordination logic; S43, calculating the efficiency and accuracy of fault removal; S44. Based on the data results, evaluate the coordination effect between the relay protections and determine the type of coordination relationship; In step S44, the coordination effect between the relay protections is evaluated through risk assessment; Key indicators reflecting the coordination effect of relay protection include: action time difference, fault removal time, false operation rate and missed detection rate; Action time difference risk indicator: ΔT risk =max(ΔT ij )-Δt th , where ΔT ij is the operating time difference between relay protection devices i and j, Δt th is the acceptable threshold of action time difference; Fault removal time risk index: T cl =T cla -T cli , where T cla is the actual fault clearing time, T cli is the ideal fault removal time; Misoperation rate risk indicators: Among them, N f is the number of false trips, N t is the total number of actions; Missed detection rate risk indicator: The above risk indicators are weighted and summed to obtain a comprehensive risk indicator: Among them, ω i is the weight of the ith risk indicator, R i is the value of the ith risk indicator, and n is the number of risk indicators.

10. A relay protection simulation system, based on a relay protection simulation method according to any one of claims 1 to 9, characterized in that: Modular protection function modules, including: longitudinal current differential protection module, distance protection module, zero sequence protection module and other protection modules, are used to simulate the protection methods in the actual power system and perform fault detection and removal on the power system; Data preprocessing module, processing the collected raw data; The deep learning model module uses CNN and LSTM to extract features and analyze time series of preprocessed data to identify and classify faults; The simulation test module sets fault points and abnormal conditions in the simulation environment, runs the model and observes the action of the protection device; The coordination relationship analysis module analyzes the coordination relationship between protection devices, identifies weak links through risk assessment, and optimizes and adjusts the relay protection system.

Images