An intelligent arc suppression and current limiting method and system based on dynamic fault prediction

Abstract

The present invention discloses an intelligent arc suppression and current limiting method and system based on dynamic fault prediction, which relates to the technical field of power systems and includes: collecting distribution network data, performing denoising and data smoothing processing; predicting the current trend according to the distribution network data, predicting the fault occurrence time and location, generating a fault prediction report, locating the fault point and quickly isolating the fault area; dynamically adjusting the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performing feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limiting. The method of the present invention not only improves the automation and intelligence level of fault handling, but also greatly shortens the response time of fault handling, and improves the operation reliability and stability of the power grid.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and particularly to an intelligent arc suppression and current limiting method and system based on dynamic fault prediction. Background Art

[0002] With the continuous development of modern power systems and the expansion of the power grid scale, the operating environment of the distribution network has become increasingly complex, and the occurrence frequency of various faults has also been increasing year by year. Traditional distribution network fault detection and handling methods are difficult to meet the requirements of the current complex power grid. In recent years, intelligent technologies have been increasingly widely applied in the distribution network, including fault prediction technologies based on artificial intelligence and machine learning. These technologies can analyze a large amount of historical data and can relatively accurately predict the occurrence time and location of faults, thereby improving the operating reliability and stability of the power grid.

[0003] Currently, common arc suppression and current limiting methods mainly include arc suppression coils with fixed parameters and static fault current limiting devices. Although these methods can alleviate the impact of faults to a certain extent, they are often difficult to achieve fast and accurate fault handling in the face of a complex and changeable power grid environment. The arc suppression coil with fixed parameters has limited effects in different fault situations and cannot be dynamically adjusted according to the actual situation. The static fault current limiting device can act quickly when a fault occurs, but its fixed working mode cannot adapt to various changes in the power grid operation and it is difficult to achieve the best current limiting effect.

[0004] There are significant deficiencies in the prior art in terms of fault prediction and handling. First, traditional fault detection methods rely on fixed thresholds and preset fault modes and are difficult to cope with diverse fault types in the distribution network. Second, existing arc suppression and current limiting devices lack real-time feedback and dynamic adjustment mechanisms and cannot be effectively adjusted according to the actual situation of faults, resulting in poor fault handling effects. In addition, the prior art usually involves manual intervention for handling after a fault occurs, increasing the response time and operation complexity of fault handling. Summary of the Invention

[0005] In view of the above problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that the prior art lacks intelligent fault prediction capabilities, and the parameters of arc suppression and current limiting devices are fixed and cannot be dynamically adjusted according to the actual fault situation.

[0007] To solve the above technical problems, the present invention provides the following technical solutions: An intelligent arc suppression and current limiting method based on dynamic fault prediction, including: collecting distribution network data, performing denoising and data smoothing processing; predicting the current trend based on the distribution network data, predicting the fault occurrence time and location, generating a fault prediction report, locating the fault point and quickly isolating the fault area; dynamically adjusting the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performing feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limiting.

[0008] As a preferred embodiment of the intelligent arc suppression and current limiting method based on dynamic fault prediction according to the present invention, wherein: the distribution network data includes the current, voltage, and temperature data of the line, the working state data of the arc suppression coil, and historical data.

[0009] As a preferred embodiment of the intelligent arc suppression and current limiting method based on dynamic fault prediction according to the present invention, wherein: the current trend prediction includes constructing a combined model of LSTM and GRU to identify and predict the current trend. The structure of the model is divided into an input layer, an LSTM layer, a GRU layer, a weight calculation layer, a weighted average layer, and an output layer. The input layer receives the distribution network data and uploads it to the LSTM layer and the GRU layer for calculation respectively to obtain the hidden state output of the LSTM and the hidden state output of the GRU ; calculating the adaptive weight parameter in the weight calculation layer , using a sliding window to calculate the mean square error within the time window , and updating it through gradient descent , expressed as:

[0010] ,

[0011] ,

[0012] wherein, represents the updated adaptive weight parameter; represents the initial adaptive weight parameter; represents the learning rate; represents the time window within the mean square error; represents the moment; represents the mean square error at time t;

[0013] At the same time, in the weighted average layer, according to the adaptive weight parameter , combine and by weighting, and output the combined result through the output layer, expressed as:

[0014] ,

[0015] Among them, represents the predicted value of the current at time ; the safety range of the current is set based on historical data , the threshold of the current change rate and , and ; represents the adaptive weight parameter at time ; when is within the range of , it indicates that the current is normal; when is outside the range of , it indicates that the current is abnormal; when the predicted value of the current change rate , it indicates that the current change rate is normal; when , it indicates that the current change rate is abnormal; when , it indicates that the current change rate reaches the fault range.

[0016] As a preferred solution of the intelligent arc suppression and current limiting method based on dynamic fault prediction described in the present invention, among them: the initial adaptive weight parameter is obtained through grid search during model training, and the value range of the initial adaptive weight parameter is set to , the initial step size is , and the number of cross-validation folds is ;

[0017] For each value, grid search is performed: initialize to 0, divide the dataset used for training into subsets, for each subset , use the remaining except subsets to train the combined model, use subset to validate the model, calculate the mean square error , accumulate to , calculate the average mean square error of all subsets, expressed as:

[0018] ,

[0019] Among them, represents the average mean square error of ; record the value of that makes the smallest, denoted as , and narrow the value range of to , and set the step size to , perform grid search on each value again and calculate , repeat the steps of grid search and narrowing the range of values until the target accuracy is reached, and output as the initial adaptive weight parameter.

[0020] As a preferred solution of the intelligent arc suppression and current limiting method based on dynamic fault prediction according to the present invention, wherein: the fault prediction report includes that when the current is normal and the current change rate is normal, it is determined that the distribution network is in a normal operation state, the existing arc suppression coil parameters are maintained and no fault prediction report is generated; when the current is normal and the current change rate is abnormal, it is determined that the distribution network is in an abnormal fluctuation state, the sampling frequency of the distribution network data is increased, and the arc suppression coil parameters are dynamically adjusted. If the duration of the abnormal fluctuation state exceeds the preset time threshold , a warning signal is sent to remind the staff to monitor manually; when the current is normal and the current change rate reaches the fault range, it is determined that the distribution network is in a potential fault state, the arc suppression coil parameters are dynamically adjusted, the location of the potential fault is reversely located according to the data source of the distribution network, a fault prediction report is generated and a warning is issued; when the current is abnormal and the current change rate is normal, it is determined that the distribution network is in a current abnormal state, the sampling frequency of the distribution network data is increased, the arc suppression coil parameters are dynamically adjusted to prevent the current from further abnormal, a warning signal is sent to remind the staff to check the load condition and prevent further deterioration; when the current is abnormal and the current change rate is abnormal, it is determined that the distribution network is in a complex abnormal state, the fault detection sensor is started to scan the complex abnormal area, the arc suppression coil parameters are dynamically adjusted and the auxiliary protection device is started to prevent the fault from expanding, a fault prediction report is generated and a warning is issued, and the dispatching staff conducts on-site confirmation and maintenance; when the current is abnormal and the current change rate reaches the fault range, it is determined that the distribution network is in a fault state, the arc suppression coil parameters are immediately adjusted to reduce the fault current, the fault detection sensor is started to locate the position where the fault occurs, the fault area is quickly isolated to prevent the fault from expanding, after confirming that the fault area has been powered off and isolated, the operation of the arc suppression equipment is stopped, a fault prediction report is generated and a warning is issued, the dispatching staff conducts fault repair, and after the staff has completed the repair and manually confirmed, the isolation measures for the fault area are withdrawn.

[0021] As a preferred solution of the intelligent arc suppression and current limiting method based on dynamic fault prediction according to the present invention, wherein: the dynamic adjustment of the arc suppression coil parameters includes that when the distribution network is in an abnormal fluctuation state, the damping characteristic of the system is increased by slightly adjusting the inductance to suppress the fluctuation, expressed as:

[0022] ,

[0023] wherein, Represents the adjusted inductance; Represents the inductance before adjustment; Represents the adjustment amount;

[0024] When the distribution network is in a potential fault state, change the connection mode of the arc suppression coil from parallel to series and adjust the inductance to quickly suppress abnormal fluctuations, expressed as:

[0025] ,

[0026] Among them, Represents the adjustment coefficient;

[0027] When the distribution network is in an abnormal current state, optimize the arc suppression effect by adjusting the ratio of inductance to resistance, expressed as:

[0028] ,

[0029] Among them, Represents the adjusted resistance; Represents the resistance before adjustment; Represents the deviation of abnormal current; Represents the normal current;

[0030] When the distribution network is in a complex abnormal state, adjust the inductance and add a capacitor to form an LC resonance circuit to absorb current fluctuations, expressed as:

[0031] ,

[0032] ,

[0033] Among them, Represents the adjusted capacitor; Represents the capacitor before adjustment; Represents the newly added capacitor value;

[0034] When the distribution network is in a fault state, change the connection mode of the arc suppression coil from parallel to series and adjust the inductance to quickly suppress abnormal fluctuations. If there is a standby arc suppression device, start it to provide additional protection, and stop the arc suppression work after the fault area is isolated, expressed as:

[0035] ,

[0036] ,

[0037] Among them, Represents the adjustment coefficient; Represents the current change rate.

[0038] As a preferred solution of the intelligent arc suppression and current limiting method based on dynamic fault prediction described in the present invention, where: the feedback control includes calculating the error between the actual value and the predicted value after adjusting the arc suppression coil parameters, expressed as:

[0039] ,

[0040] where, represents the error at time ; represents the predicted value at time ; represents the actual measured value at time ; when the error increases , it indicates that the current adjustment amplitude is insufficient, and increase , expressed as:

[0041] ,

[0042] when the error decreases , it indicates that the current adjustment amplitude is too large, and decrease , expressed as:

[0043] ,

[0044] where, represents the new adjustment amount; represents the adjustment coefficient.

[0045] An intelligent arc suppression and current limiting system based on dynamic fault prediction adopting any of the methods described in the present invention, where: a collection module, which collects distribution network data, performs denoising and data smoothing processing, and uploads the processed data to the detection module; a detection module, which constructs a combined model of LSTM and GRU to identify and predict the current trend, optimizes the prediction accuracy by dynamically adjusting the weights of the weighted average layer, generates a fault prediction report according to the prediction result, locates the fault point and quickly isolates the fault area; an arc suppression module, which dynamically adjusts the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performs feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limiting.

[0046] A computer device, including: a memory and a processor; the memory stores a computer program, including: when the processor executes the computer program, the steps of any of the methods described in the present invention are implemented.

[0047] A computer-readable storage medium, on which a computer program is stored, including: when the computer program is executed by a processor, the steps of any of the methods described in the present invention are implemented.

[0048] Advantages of the present invention: The method of the present invention collects distribution network data, performs denoising and data smoothing processing, uses a current trend prediction model to predict the occurrence time and location of faults, generates a fault prediction report, locates the fault point and quickly isolates the fault area. At the same time, according to the real-time monitoring data and the fault prediction report, the parameters of the arc suppression coil are dynamically adjusted, and the arc suppression effect is monitored in real time by using a feedback control mechanism to achieve rapid arc suppression and current limitation. It not only improves the automation and intelligence level of fault handling, but also greatly shortens the response time of fault handling and improves the operation reliability and stability of the power grid. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings based on these drawings without creative efforts.

[0050] Figure 1 FIG. is an overall flowchart of an intelligent arc suppression and current limitation method based on dynamic fault prediction provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In order to make the above objects, features, and advantages of the present invention more obvious and understandable, the specific embodiments of the present invention will be described in detail below with reference to the drawings of the specification. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope of protection of the present invention.

[0052] Embodiment 1, referring to Figure 1 , which is an embodiment of the present invention, provides an intelligent arc suppression and current limitation method based on dynamic fault prediction, including:

[0053] S1: Collect distribution network data, perform denoising and data smoothing processing.

[0054] Further, after the system is started, historical fault data and the working state data of the arc suppression coil are loaded, the sensor network is initialized to ensure that all monitoring devices are working properly, and connected to the control center.

[0055] The loading of historical fault data is the basis for the system to perform intelligent fault prediction. By analyzing the past fault data, the fault prediction model can identify common fault patterns and rules, thereby improving the accuracy of fault prediction.

[0056] The working state data of the arc suppression coil is used to understand the historical operation conditions and performance indicators of the equipment, ensuring that the actual working ability and state of the arc suppression coil can be taken into account when predicting faults, which is of great significance for adjusting the parameters of the arc suppression coil and optimizing the arc suppression effect. When adjusting the arc suppression coil according to the fault prediction results, the adjustment amount and the basic amount are obtained based on the working state data of the arc suppression coil.

[0057] Use the sensor network to monitor parameters such as current, voltage, and temperature in real time, transmit the real-time monitoring data to the control center for preliminary processing and storage, and preprocess the collected data. Denoising is the first step in data preprocessing, aiming to improve the accuracy and reliability of the data.

[0058] Monitoring data such as current, voltage, and temperature may be affected by various noises and interferences during the acquisition process, such as electromagnetic interference, equipment failures, environmental changes, etc. These noises and outliers will affect the accuracy of the fault prediction model. Therefore, denoising must be carried out. The Kalman filter is used to process the data online in real time, and the dynamic model and measurement model of the system are used to predict and correct the monitoring data.

[0059] Smoothing is the second step in data preprocessing, aiming to eliminate the short-term fluctuations in the data, highlight the long-term trends, and make the data more stable and coherent. Different weights are assigned to the data within the window through weighted moving average. Usually, the data closer to the center point has a larger weight, which can better retain the local characteristics of the data and achieve the smoothing effect at the same time. The initial value of the size of the sliding window is adjusted according to the fault prediction report. If a fault risk or current fluctuation is detected, it is reduced to increase the monitoring frequency.

[0060] S2: Perform current trend prediction based on the distribution network data, predict the time and location of the fault occurrence, generate a fault prediction report, locate the fault point, and quickly isolate the fault area.

[0061] Furthermore, in practical applications, the fault prediction of the power system considers both long-term and short-term data characteristics. LSTM and GRU are two commonly used variants of the recurrent neural network (RNN). LSTM is a long short-term memory network, which performs well in processing long sequence data and can better retain long-term memories; GRU is a gated recurrent unit, which performs well in processing short sequence data and has higher computational efficiency.

[0062] Using LSTM or GRU alone may not be able to comprehensively capture all the characteristics in the data. Therefore, combining the advantages of LSTM and GRU can analyze the data more comprehensively and provide more accurate predictions.

[0063] The main part of the combination of LSTM and GRU is their hidden state outputs (i.e., their respective final hidden states). In traditional LSTM and GRU structures, their respective hidden states are used to predict the next output. In the combined neural network, the hidden state outputs of LSTM and GRU are calculated separately, and these two hidden states are weighted and averaged according to an adaptive weight parameter to obtain the final prediction result.

[0064] The structure of constructing the combined model of LSTM and GRU is divided into an input layer, an LSTM layer, a GRU layer, a weight calculation layer, a weighted average layer, and an output layer. The input layer receives the distribution network data and uploads it to the LSTM layer and the GRU layer respectively for calculation to obtain the hidden state output of LSTM and the hidden state output of GRU .

[0065] The adaptive weight parameter calculated in the weight calculation layer , uses a sliding window to calculate the mean squared error within the time window and is updated through gradient descent , expressed as:

[0066] ,

[0067] ,

[0068] where, represents the updated adaptive weight parameter; represents the initial adaptive weight parameter; represents the learning rate; represents the time window within the mean squared error; represents the time instant; represents the mean squared error at time t.

[0069] It should be noted that since the hidden state outputs of LSTM and GRU are both in vector form, they can be combined through weighted averaging without additional deformation adjustment of the output form, and there is no need to bear the risk of increased computational complexity and additional errors caused by data deformation. The prediction accuracy can be improved through simple weighted combination.

[0070] Over time, the environment and operating conditions in the distribution network may change, resulting in a decline in the performance of the fault prediction model. To maintain the accuracy and robustness of the model, the initial weights are calculated during model training, but fixed values are not used in actual applications. Instead, the values are dynamically adjusted through real-time monitoring to optimize the performance of the model.

[0071] Meanwhile, in the weighted average layer, according to the adaptive weight parameter , combine and by weighted combination, and output the combination result through the output layer, expressed as:

[0072] ,

[0073] wherein, represents the predicted value of the current at time ; set the safety range of the current based on historical data , the threshold of the current change rate and , and ; represents the adaptive weight parameter at time .

[0074] When is within the range of , it indicates that the current is normal; when is outside the range of , it indicates that the current is abnormal.

[0075] When the predicted value of the current change rate , it indicates that the current change rate is normal; when , it indicates that the current change rate is abnormal; when , it indicates that the current change rate reaches the fault range.

[0076] Furthermore, during the model training process, the optimal parameters are calculated through grid search and cross-validation. Set the initial value range to be , the initial step size to be , and the number of cross-validation folds to be .

[0077] For each value, perform grid search:

[0078] Initialize to 0, divide the dataset used for training into subsets. For each subset , use the remaining subsets except to train the combination model, use subset to validate the model, calculate the mean square error , accumulate to , and calculate the average mean square error of all subsets, expressed as:

[0079] ,

[0080] Among them, represents the average mean square error; record the value that minimizes and denote it as Reduce the value range to and set the step size to Then, perform grid search on each value again and calculate Repeat the grid search and the step of reducing the value range until reaches the target accuracy, and output as the initial adaptive weight parameter.

[0081] Reduce the range according to the search results of the previous round and refine the step size for fall calculation, and a more accurate can be found as the initial adaptive weight parameter. Dynamically adjusting the range and step size can reduce unnecessary calculations, improve the search efficiency, ensure that the parameter optimization is accurate to several decimal places, and enhance the model performance.

[0082] It should be noted that is recalculated and corrected at certain time intervals or when specific conditions are met to adapt to environmental changes. For example, grid search and cross-validation can be performed again at regular intervals (such as monthly) or when the prediction performance drops to a certain threshold

[0083] Furthermore, a fault prediction report is generated based on the prediction results. When the current is normal and the current change rate is normal, it is determined that the distribution network is in a normal operating state, and the existing arc suppression coil parameters are maintained without generating a fault prediction report. At this time, the system is in a stable state, ensuring that the system remains stable under normal conditions, without unnecessary adjustments, and avoiding introducing new disturbances.

[0084] When the current is normal and the current change rate is abnormal, it is determined that the distribution network is in an abnormal fluctuation state. Increase the sampling frequency of the distribution network data and dynamically adjust the arc suppression coil parameters. If the duration of the abnormal fluctuation state exceeds the preset time threshold , a warning signal is sent to remind the staff for manual monitoring.

[0085] When the current is normal and the current change rate reaches the fault range, it is determined that the distribution network is in a potential fault state. Dynamically adjust the arc suppression coil parameters, reverse-locate the occurrence location of the potential fault according to the distribution network data source, generate a fault prediction report and issue a warning.

[0086] When the current is abnormal and the current change rate is normal, it is determined that the distribution network is in an abnormal current state. Increase the sampling frequency of the distribution network data, dynamically adjust the arc suppression coil parameters to prevent the current from further abnormal, send out a warning signal to remind the staff to check the load situation, and prevent further deterioration.

[0087] When the current is abnormal and the current change rate is abnormal, it is determined that the distribution network is in a complex abnormal state. Start the fault detection sensor to scan the complex abnormal area, dynamically adjust the arc suppression coil parameters and start the auxiliary protection device to prevent the fault from expanding, generate a fault prediction report and give a warning, and dispatch the staff to conduct on-site confirmation and maintenance.

[0088] When the current is abnormal and the current change rate reaches the fault range, it is determined that the distribution network is in a fault state. Immediately adjust the arc suppression coil parameters to reduce the fault current, start the fault detection sensor to locate the fault location, quickly isolate the fault area to prevent the fault from expanding, stop the operation of the arc suppression equipment after confirming that the fault area has been powered off and isolated, generate a fault prediction report and give a warning, dispatch the staff to repair the fault, and withdraw the isolation measures for the fault area after the staff has completed the repair and manually confirmed.

[0089] S3: According to the distribution network data and the fault prediction report, dynamically adjust the parameters of the arc suppression coil, and feedback control the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limitation.

[0090] Further, dynamically adjusting the parameters of the arc suppression coil includes that when the distribution network is in an abnormal fluctuation state, slightly adjust the inductance to increase the damping characteristic of the system and suppress the fluctuation, which is expressed as:

[0091] ,

[0092] Among them, represents the adjusted inductance; represents the inductance before adjustment; represents the adjustment amount.

[0093] When the distribution network is in a potential fault state, change the connection mode of the arc suppression coil from parallel to series and adjust the inductance to quickly suppress the abnormal fluctuation, which is expressed as:

[0094] ,

[0095] Among them, represents the adjustment coefficient.

[0096] When the distribution network is in an abnormal current state, optimize the arc suppression effect by adjusting the ratio of the inductance to the resistance, which is expressed as:

[0097] ,

[0098] Among them, represents the adjusted resistance; represents the resistance before adjustment; represents the deviation of abnormal current; represents the normal current.

[0099] When the distribution network is in a complex abnormal state, adjust the inductance and add a capacitor to form an LC resonance circuit to absorb current fluctuations, which is expressed as:

[0100] ,

[0101] ,

[0102] Among them, represents the adjusted capacitor; represents the capacitor before adjustment; represents the newly added capacitor value.

[0103] Usually, the additional capacitor CCC is not normally connected to the circuit to avoid unnecessary current resonance under normal operating conditions. In a complex abnormal state, the capacitor CCC is connected to the circuit through a switching switch to form an LC resonance circuit. The resonance frequency of the LC resonance circuit is expressed as:

[0104] ,

[0105] Among them, L represents the inductance; C represents the capacitance; when the current fluctuation frequency is close to the resonance frequency, the LC resonance circuit can effectively absorb the high-frequency components of the current and reduce the current fluctuation.

[0106] When the distribution network is in a fault state, change the connection method of the arc suppression coil from parallel to series and adjust the inductance to quickly suppress abnormal fluctuations. If there is a standby arc suppression device, start it to provide additional protection. Stop the arc suppression work after the fault area is isolated, which is expressed as:

[0107] ,

[0108] ,

[0109] Among them, represents the adjustment coefficient; represents the current change rate.

[0110] It should be noted that the connection method of the arc suppression coil to the circuit is divided into two types: series and parallel. Under normal circumstances, the arc suppression coil is usually connected in parallel to the distribution system, so that when a fault current appears in the system, it can quickly share the current and reduce the probability of arc occurrence.

[0111] In certain situations, such as when a higher inductance is required to limit the rate of fault current, a series connection can be selected. However, this method increases the inductance of the system during normal operation, which may affect the system performance.

[0112] Therefore, when a fault occurs or the power grid fluctuates significantly, the arc suppression coil is changed to a series connection to provide a higher inductance. After the fluctuation becomes smaller and returns to the normal range, it is switched back to a parallel connection to ensure the system performance.

[0113] Furthermore, the feedback control includes calculating the error between the actual value and the predicted value after adjusting the parameters of the arc suppression coil, which is expressed as:

[0114] ,

[0115] where, represents the error at time ; represents the predicted value at time ; represents the actual measured value at time ;

[0116] When the error increases , it indicates that the current adjustment amplitude is insufficient, and is increased, which is expressed as:

[0117] ,

[0118] When the error decreases , it indicates that the current adjustment amplitude is too large, and is decreased, which is expressed as:

[0119] ,

[0120] where, represents the new adjustment amount; represents the adjustment coefficient.

[0121] On the other hand, this embodiment also provides an intelligent arc suppression and current limiting system based on dynamic fault prediction, which includes: a data acquisition module that acquires the data of the distribution network, performs denoising and data smoothing processing, and uploads the processed data to the detection module; a detection module that constructs a combined model of LSTM and GRU to identify and predict the current trend, optimizes the prediction accuracy by dynamically adjusting the weights of the weighted average layer, generates a fault prediction report according to the prediction result, locates the fault point and quickly isolates the fault area; an arc suppression module that dynamically adjusts the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performs feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve fast arc suppression and current limiting.

[0122] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs.

[0123] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in combination with an instruction execution system, apparatus, or device.

[0124] More specific examples (non-exhaustive list) of computer-readable media include the following: electrical connection parts with one or more wirings (electronic devices), portable computer disk cartridges (magnetic devices), random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memories), fiber optic devices, and portable compact disc read-only memories (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other media, then editing, interpreting, or otherwise processing it as necessary, and then storing it in a computer memory.

[0125] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), etc.

[0126] Embodiment 2. Hereinafter, an embodiment of the present invention provides an intelligent arc suppression and current limiting method based on dynamic fault prediction. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through simulation experiments.

[0127] Use PSCAD / EMTDC to establish a simulation model of the distribution network, including monitoring nodes for current, voltage, and temperature, and build a dynamic adjustment model of the arc suppression coil in Simulink.

[0128] Run multiple simulations in the simulation model to generate current, voltage, and temperature data under different working conditions, and use Python to perform denoising processing (using Kalman filtering) and data smoothing processing (using sliding window technology) on the simulation-generated data.

[0129] Use deep learning libraries in Python (such as TensorFlow or PyTorch) to build a fault prediction model combining LSTM and GRU, and use the preprocessed simulation data as training data to train the fault prediction model.

[0130] Run the trained fault prediction model, use the dynamic weights mentioned in the present invention to weight and combine the outputs of the hidden layer, predict the current trend and the time and location of fault occurrence, dynamically adjust the parameters of the arc suppression coil according to the prediction results, and use a feedback control algorithm to monitor the arc suppression effect in real time. The experimental data is shown in the following table.

[0131] Table 1 Experimental data table

[0132] ,

[0133] It can be seen from the tabular data that in Experiment 2 and Experiment 5, the fault prediction model accurately predicted the fault occurrence time and location. The predicted fault times were 130 seconds and 135 seconds, corresponding to Node 5 and Node 2 respectively. The actual fault occurrence time was very close to the predicted time, with an error within ±5 seconds, and the prediction of the fault location was also very accurate, verifying that the fault prediction model combining LSTM and GRU performs well in current trend prediction and fault location.

[0134] Based on the prediction results, dynamic adjustment of the inductance and resistance is carried out. For example, in Experiment 2, the inductance adjustment amount is 12 mH and the resistance adjustment amount is 0.6 Ω. The actual current and voltage are effectively controlled after the fault occurs. The current fluctuation decreases and the voltage is maintained within the normal range, and the arc extinguishing effect is obvious.

[0135] In Experiment 2, after the fault occurred, the current decreased from 52 A to 48 A and the voltage decreased from 225 V to 215 V, indicating that the dynamic adjustment of the arc suppression coil effectively controlled the fault impact. In Experiment 5, after the fault occurred, the current decreased from 51 A to 45 A and the voltage decreased from 224 V to 210 V, and the dynamic adjustment was also effective.

[0136] In summary, the method of the present invention combines an advanced deep learning fault prediction model and a method of dynamically adjusting the parameters of the arc suppression coil, realizes accurate prediction and rapid response to distribution network faults, and improves the reliability and safety of the power grid. The experimental data fully verifies the effectiveness and practicability of the present invention under different working conditions.

[0137] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. An intelligent arc suppression and current limiting method based on dynamic fault prediction, characterized in that, Including: Collecting distribution network data and performing denoising and data smoothing processing; Predicting the current trend based on the distribution network data, predicting the time and location of faults, generating a fault prediction report, locating the fault point and quickly isolating the fault area; The above-mentioned current trend prediction includes constructing a combined model of LSTM and GRU to identify and predict the current trend. The structure of the model is divided into an input layer, an LSTM layer, a GRU layer, a weight calculation layer, a weighted average layer, and an output layer. The input layer receives the distribution network data and uploads it to the LSTM layer and the GRU layer respectively for calculation to obtain the hidden state output of the LSTM and the hidden state output of the GRU Calculating the adaptive weight parameter α in the weight calculation layer, calculating the mean square error within the time window W using a sliding window, and updating α through gradient descent, expressed as: Among them, α represents the updated adaptive weight parameter; α c represents the initial adaptive weight parameter; η represents the learning rate; MSE W represents the mean square error within the time window W; t represents the moment; represents the mean square error at the moment t; Meanwhile, in the weighted average layer, according to the adaptive weight parameter α, and are weighted and combined, and the combined result is output through the output layer, expressed as: Among them, y t represents the predicted current value at time t; based on historical data, a safety range of current (I min , I max ), thresholds ΔI1 and ΔI2 of the current change rate are set, and ΔI1 < ΔI2; α t represents the adaptive weight parameter at time t; When y t is within the range of (I min , I max ), it indicates that the current is normal; When y t Outside the range of (I min , I max ), it indicates an abnormal current; When the predicted value Δy of the current change rate t When <ΔI1, it indicates that the current change rate is normal; When ΔI1 ≤ Δy t When <ΔI2, it indicates an abnormal current change rate; When Δy t ≥ΔI2, it indicates that the current change rate reaches the fault range; Dynamically adjusting the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performing feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limitation.

2. The intelligent arc suppression and current limiting method based on dynamic fault prediction according to claim 1, wherein: The distribution network data includes the current, voltage, and temperature data of the line, the working state data of the arc suppression coil, and historical data.

3. The intelligent arc suppression and current limiting method based on dynamic fault prediction according to claim 2, characterized in that: The initial adaptive weight parameter α c is obtained by grid search during model training. Set the initial adaptive weight parameter α c to have a value range of (0, 1), an initial step size of b, and a cross-validation fold number of k; For each α c value, perform a grid search: Initialization Set it to 0. Divide the dataset used for training into k subsets. For each subset i, use the remaining k - 1 subsets except i to train the combined model, use subset i to validate the model, and calculate the mean squared error Add to Accumulate it to. Calculate the average mean squared error of all subsets, expressed as: Among them, represents the mean square error of α c ; record the α that minimizes c , and denote it as Reduce the value range of α c to Set the step size to 0.2b, and perform grid search and calculation for each α c value again Repeat the grid search and the step of reducing the value range of α c until α c reaches the target accuracy, and output α c as the initial adaptive weight parameter.

4. The intelligent arc suppression and current limiting method based on dynamic fault prediction according to claim 3, characterized in that: The fault prediction report includes that when the current is normal and the current change rate is normal, it is determined that the distribution network is in a normal operating state, maintaining the existing arc suppression coil parameters and not generating a fault prediction report; When the current is normal and the current change rate is abnormal, it is determined that the distribution network is in an abnormal fluctuation state, increasing the sampling frequency of the distribution network data, dynamically adjusting the parameters of the arc suppression coil. If the duration of the abnormal fluctuation state exceeds the preset time threshold T, a warning signal is sent to remind the staff for manual monitoring; When the current is normal and the current change rate reaches the fault range, it is determined that the distribution network is in a potential fault state, dynamically adjusting the parameters of the arc suppression coil, reversely locating the occurrence location of the potential fault according to the source of the distribution network data, generating a fault prediction report and giving a warning; When the current is abnormal and the current change rate is normal, it is determined that the distribution network is in a current abnormal state, increasing the sampling frequency of the distribution network data, dynamically adjusting the parameters of the arc suppression coil to prevent the current from further abnormal, sending a warning signal to remind the staff to check the load situation to prevent further deterioration; When the current is abnormal and the current change rate is abnormal, it is determined that the distribution network is in a complex abnormal state, starting the fault detection sensor to scan the complex abnormal area, dynamically adjusting the parameters of the arc suppression coil and starting the auxiliary protection device to prevent the fault from expanding, generating a fault prediction report and giving a warning, and the dispatching staff conducts on-site confirmation and maintenance; When the current is abnormal and the current change rate reaches the fault range, it is determined that the distribution network is in a fault state, immediately adjusting the parameters of the arc suppression coil to reduce the fault current, starting the fault detection sensor to locate the position where the fault occurs, quickly isolating the fault area to prevent the fault from expanding, stopping the operation of the arc suppression equipment after confirming that the fault area has been powered off and isolated, generating a fault prediction report and giving a warning, and the dispatching staff conducts fault repair. After the staff has completed the repair and manually confirmed, the isolation measures for the fault area are withdrawn.

5. The intelligent arc suppression and current limiting method based on dynamic fault prediction according to claim 4, wherein: The dynamic adjustment of the parameters of the arc suppression coil includes that when the distribution network is in an abnormal fluctuation state, slightly adjusting the inductance to increase the damping characteristic of the system to suppress the fluctuation, expressed as: L n = L c + ΔL Among them, L n represents the adjusted inductance; L c represents the inductance before adjustment; ΔL represents the adjustment amount; When the distribution network is in a potential fault state, changing the connection mode of the arc suppression coil from parallel to series and adjusting the inductance to quickly suppress the abnormal fluctuation, expressed as: L n = L c + kΔL where k represents the adjustment coefficient; When the distribution network is in a current abnormal state, optimizing the arc suppression effect by adjusting the ratio of the inductance to the resistance, expressed as: Among them, R n represents the adjusted resistance; R c represents the resistance before adjustment; ΔI represents the deviation of abnormal current; I no represents the normal current; When the distribution network is in a complex abnormal state, adjust the inductance and increase the capacitance to form an LC resonance circuit to absorb current fluctuations, expressed as: L n = L c + ΔL C n = C c + ΔC Among them, C n represents the adjusted capacitance; C c represents the capacitance before adjustment; ΔC represents the newly added capacitance value; When the distribution network is in a fault state, change the connection mode of the arc suppression coil from parallel to series, and adjust the inductance to quickly suppress abnormal fluctuations. If there is a standby arc suppression device, start it to provide additional protection. Stop the arc suppression work after the fault area is isolated, expressed as: L n = L c + kΔL Among them, θ represents the adjustment coefficient; represents the current change rate.

6. The intelligent arc suppression and current limiting method based on dynamic fault prediction according to claim 5, characterized in that: The feedback control includes calculating the error between the actual value and the predicted value after adjusting the parameters of the arc suppression coil, expressed as: Among them, e t represents the error of time t; represents the predicted value of time t; y t represents the actual measured value of time t; When the error increases by e t > e t-1 it indicates that the current adjustment amplitude is insufficient. Increase ΔL, which is expressed as: ΔL new = ΔL·(1 + β) When the error increases by e t <e t-1 it means that the current adjustment amplitude is too large. Decrease ΔL, which is expressed as: ΔL new = ΔL·(1 - β) Among them, ΔL new represents the new adjustment amount; β represents the adjustment coefficient.

7. An intelligent arc suppression and current limiting system based on dynamic fault prediction using the method according to any one of claims 1 to 6, characterized in that, Including, A collection module that collects distribution network data, performs denoising and data smoothing processing, and uploads the processed data to the detection module; A detection module that constructs a combined model of LSTM and GRU to identify and predict the current trend, optimizes the prediction accuracy by dynamically adjusting the weights of the weighted average layer, generates a fault prediction report according to the prediction result, locates the fault point and quickly isolates the fault area; An arc suppression module that dynamically adjusts the parameters of the arc suppression coil according to the distribution network data and the fault prediction report, and performs feedback control on the parameters of the arc suppression coil according to the real-time monitored arc suppression effect to achieve rapid arc suppression and current limitation.

8. A computer device, comprising: A memory and a processor; The memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the intelligent arc suppression and current limitation method based on dynamic fault prediction according to any one of claims 1-6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, the steps of the intelligent arc suppression and current limitation method based on dynamic fault prediction according to any one of claims 1-6 are implemented.

Images