A relay circuit short circuit protection method and system
By establishing a circuit state feature model and using a recurrent neural network to identify current and voltage waveform offsets, dynamically adjusting the threshold and response mechanism, the shortcomings of the existing technology relay loop short-circuit protection method under non-standard conditions are solved, and the further improvement is achieved Efficient and safe protection.
Patent Information
- Application Number
- CN202411096390.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-12
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-08-12
AI Technical Summary
The existing relay loop short-circuit protection methods lack adaptability to current fluctuations and complex circuit behavior, resulting in insufficient or excessive protection response under non-standard operating conditions, and the inability to respond to rapidly changing load requirements in a timely manner, increasing the vulnerability of protection. Especially in modern industries and power grids, there is a risk of equipment damage or safety incidents.
Using the output data based on the current sensor, the correlation model between morphological characteristics and circuit state is established by separating the mutation points and peaks, and the time series analysis is used for cyclic neural network, identify the offsets of current and voltage waveforms, and extract the change trend characteristics. And the model weight and deviation values are adjusted through cross-verification technology, and the threshold and response mechanism are dynamically adjusted to achieve real-time monitoring and automatic circuit breaking protection.
Improves the accuracy and adaptability of protection responses, reduces the risks caused by equipment failures or operating errors, ensures stable operation and safety of the power system, and can self-adjust to adapt to new or changing grid conditions.
Smart Images

Figure CN118630694B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of short circuit protection, and in particular to a relay circuit short circuit protection method and system. Background Art
[0002] Short-circuit protection is a vital technology in the field of power engineering. Its main purpose is to prevent current overload from causing circuit damage or fire. It is used to detect abnormal current flow in the circuit and quickly cut off the power supply to prevent further damage. Short-circuit protection includes fuses, circuit breakers and microprocessor-based protection devices. When a short circuit or overload condition is detected, it can automatically cut off the current to protect the circuit from further physical damage. This technology is not only used in power systems in residential and commercial buildings, but also widely used in industrial equipment and high-voltage power grids.
[0003] Among them, the relay circuit short-circuit protection method focuses on achieving effective protection of relay circuits in power systems, especially for short-circuit faults. Relay circuits are usually used in power systems as part of control circuits to drive high-power loads. This protection method involves the use of specific hardware and software technologies to monitor and diagnose short-circuit conditions in circuits, so as to respond quickly when a fault occurs, cut off the faulty part, prevent the fault from spreading and equipment damage, and is used to ensure the stability of the power system and reduce unexpected risks caused by circuit problems.
[0004] Existing protection methods lack adaptability to current fluctuations and complex circuit behaviors, resulting in insufficient or excessive protection response under non-standard operating conditions. This deficiency is particularly evident in modern industrial and grid operations that are highly dependent on stable power sources. Fixed thresholds make it impossible to respond to rapidly changing load demands in a timely manner, increasing the vulnerability of protection. For example, when the grid suddenly increases in load, outdated protection mechanisms fail to effectively disconnect, causing equipment damage or more serious safety incidents. Traditional methods show insufficient adjustments when responding to new technologies or integrating renewable energy, resulting in increased maintenance costs and shortened equipment life. Summary of the invention
[0005] The purpose of the present invention is to solve the shortcomings in the prior art and to propose a relay circuit short-circuit protection method and system.
[0006] In order to achieve the above object, the present invention adopts the following technical solution: a relay circuit short circuit protection method, comprising the following steps:
[0007] S1: Based on the output data of the current sensor, separate the mutation points and peak values in the current and voltage data, establish the association model between the morphological characteristics and the circuit state, formulate the data classification standard, and generate the circuit state feature set;
[0008] S2: Based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of the current and voltage waveforms, extract the change trend characteristics, and identify and classify the feature patterns to generate an abnormal waveform feature set;
[0009] S3: Based on the abnormal waveform feature set, the cross-validation technology is used to adjust the model weight and deviation value through current data feedback, and the stability and accuracy of the model are verified by repeated experiments to generate a short circuit prediction model;
[0010] S4: According to the short circuit prediction model, using percentiles, combined with historical waveform data, and through Fourier transform, the threshold for responding to sudden current increase is calculated, and according to the overcurrent protection requirements, the threshold is dynamically adjusted and the response mechanism is set to generate a dynamic threshold configuration;
[0011] S5: Based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism, records the status of the protection action, and generates circuit breaker protection status information;
[0012] S6: Based on the circuit breaker protection status information, collect the relay circuit performance data after executing the protection action, evaluate the response effect, and refine and adjust the short circuit prediction model and threshold to generate a protection mechanism optimization result.
[0013] As a further solution of the present invention, the circuit state feature set includes voltage extreme values, current peak values, and sampling time intervals; the abnormal waveform feature set includes waveform deviation, change trend, and classification indicators; the short circuit prediction model includes parameter optimization records and verification accuracy results; the dynamic threshold configuration includes real-time adjustment logic, historical threshold data, and emergency response levels; the circuit breaker protection state includes activation delay, trigger threshold, and event log; and the protection mechanism optimization results include performance improvement indicators, adjustment plans, and efficiency evaluation results.
[0014] As a further solution of the present invention, based on the output data of the current sensor, the mutation points and peak values in the current and voltage data are separated, a correlation model between the morphological characteristics and the circuit state is established, and a data classification standard is formulated. The steps of generating the circuit state feature set are specifically as follows:
[0015] S101: Based on the output data of the current sensor, by real-time monitoring of current and voltage readings, marking the mutation points and peak values of the data, and recording the timestamp and amplitude of each mutation point, to obtain a mutation point and peak value data set;
[0016] S102: Based on the mutation point and peak value data set, analyzing the occurrence pattern of each mutation point and peak value in the data set, constructing a corresponding relationship between the data point and the circuit working state, and obtaining a primary circuit state model by comparing the data patterns under the difference state;
[0017] S103: Based on the primary circuit state model, data feature classification processing is performed, the data is sorted and encoded according to the circuit state, and the classified data is integrated to obtain a circuit state feature set.
[0018] As a further solution of the present invention, based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of the current and voltage waveforms, extract the change trend characteristics, and identify and classify the characteristic patterns. The steps of generating an abnormal waveform feature set are specifically as follows:
[0019] S201: Based on the circuit state feature set, using a recurrent neural network, analyzing the data waveforms of current and voltage, marking the offset point of each data waveform, and recording the amplitude and frequency of the offset point, analyzing the offset rule, and obtaining a waveform offset data set;
[0020] S202: extracting the rising or falling trend of each waveform based on the waveform offset data set, classifying the changing trends, and recording the waveform features in each classification to obtain a primary trend feature set;
[0021] S203: Based on the primary trend feature set, iteratively refine the features, group the waveform features according to functionality or abnormality, organize the groups into standard classifications, and obtain an abnormal waveform feature set.
[0022] As a further solution of the present invention, the recurrent neural network is according to the formula: Calculate the hidden state at each time step, identify and mark the offset points, and generate the amplitude and frequency data of the offset points, where is the time step The hidden state of is the time step The hidden state of is the time step The input data, is the weight matrix from hidden state to hidden state, is the weight matrix input to the hidden state, is the deviation term, is the activation function, , and are the weight adjustment coefficients for hidden state, input data, and context state, respectively. is the time step The context state, is the weight matrix from context state to hidden state.
[0023] As a further solution of the present invention, based on the abnormal waveform feature set, cross-validation technology is used to adjust the model weights and deviation values through current data feedback, and repeated experiments are performed to verify the stability and accuracy of the model. The steps of generating a short circuit prediction model are specifically as follows:
[0024] S301: Based on the abnormal waveform feature set, multiple groups of verification data sets are set to perform model testing, the impact of weight adjustment on the test results is analyzed, the model performance of each group of data is recorded, and multiple rounds of model performance verification are performed to generate a preliminary model data set;
[0025] S302: Based on the preliminary model data set, analyze the stability and accuracy of multiple test results, identify the best performing weight and bias value configuration, adjust the model parameters, optimize the accuracy and response speed of the model, and obtain an optimized model data set;
[0026] S303: Based on the optimized model data set, repeatedly perform multiple rounds of testing to evaluate the performance of the model under various conditions, determine the performance consistency of the model on the difference test set, and generate a short circuit prediction model.
[0027] As a further solution of the present invention, according to the short circuit prediction model, using percentiles, combined with historical waveform data, through Fourier transform, the threshold for responding to sudden current increase is calculated, and according to the overcurrent protection requirements, the threshold is dynamically adjusted and the response mechanism is set. The steps of generating a dynamic threshold configuration are specifically as follows:
[0028] S401: Based on the short circuit prediction model and in combination with historical waveform data, a preliminary threshold is set by calculating the percentile of the historical data, and the performance of the threshold under the difference waveform is analyzed to generate an initial threshold set;
[0029] S402: Based on the initial threshold set, dynamically adjust the threshold according to the current increase data monitored in real time through Fourier transform, record the parameters and results of each adjustment, analyze the impact on the current protection effect, optimize the threshold adjustment process, and obtain the adjusted threshold set;
[0030] S403: Based on the adjusted threshold set, a current surge response mechanism is set, and by simulating a differential current surge situation, a real-time response effect of the threshold is tested, the effectiveness of the threshold setting and the response mechanism is verified, and a dynamic threshold configuration is generated.
[0031] As a further solution of the present invention, based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism and records the status of the protection action. The steps of generating the circuit breaker protection status information are specifically as follows:
[0032] S501: Based on the dynamic threshold configuration, continuously monitor the newly received current and voltage data, if the detection data exceeds the set threshold, activate the protection mechanism of the circuit breaker, and record the triggering time and environmental parameters of the protection action, and generate a protection event record;
[0033] S502: Based on the protection event record, analyze the time point, environmental conditions and threshold of each circuit breaker triggering, evaluate the reaction speed and triggering accuracy of the protection mechanism, and generate a protection effect analysis result;
[0034] S503: Based on the protection effect analysis result, the data of the difference protection events are summarized and compared, the frequency, status and environmental response of the events are recorded, and the circuit breaker protection status information is generated.
[0035] As a further solution of the present invention, based on the circuit breaker protection state information, the relay circuit performance data after the protection action is executed is collected, the response effect is evaluated, and the short circuit prediction model and threshold are refined and adjusted to generate the protection mechanism optimization result. Specifically, the steps are as follows:
[0036] S601: Based on the circuit breaker protection state information, collect the relay circuit performance data after the protection action, analyze the performance of the relay circuit under different working conditions, evaluate the stability and response efficiency, and generate a relay performance analysis result;
[0037] S602: Based on the relay performance analysis result, analyze the impact of the protection action on the relay circuit performance, adjust the parameters of the short circuit prediction model and the threshold, and continuously monitor and test the effectiveness of the adjustment to generate a protection mechanism adjustment configuration;
[0038] S603: Based on the protection mechanism adjustment configuration, the short circuit prediction model and threshold settings are adjusted, multiple rounds of testing are performed to verify the adaptability and accuracy of the adjusted model and threshold under various conditions, and generate protection mechanism optimization results.
[0039] A relay circuit short circuit protection system, the relay circuit short circuit protection system is used to execute the above relay circuit short circuit protection method, the system comprises:
[0040] The current and voltage data analysis module samples the current and voltage data based on the output data of the current sensor, separates the mutation point and the peak value, and establishes the corresponding relationship with the circuit state by analyzing the morphological characteristics of the current and voltage to generate a circuit state feature set;
[0041] The time series analysis module performs time series analysis on the current and voltage waveforms based on the circuit state feature set, performs offset identification, extracts trend features of waveform changes, identifies and classifies the features, and generates an abnormal waveform feature set;
[0042] The model optimization and verification module performs model parameter optimization based on the abnormal waveform feature set, uses cross-validation technology to adjust model weights and deviation values, verifies the stability and accuracy of the model through data feedback and repeated tests, and generates a short circuit prediction model;
[0043] The threshold dynamic configuration module calculates the threshold of the burst current by using percentiles and combining historical waveform data according to the analysis results of the short circuit prediction model, dynamically adjusts the threshold and sets the response mechanism according to the protection requirements, and generates a dynamic threshold configuration;
[0044] The real-time monitoring response module continuously analyzes the newly received current and voltage data based on the dynamic threshold configuration. When the data exceeds the set threshold, the circuit breaker protection mechanism is triggered, and the performance data of the relay circuit after protection is collected to evaluate the effect, refine and adjust the short circuit prediction model and threshold configuration, and generate the protection mechanism optimization result.
[0045] Compared with the prior art, the advantages and positive effects of the present invention are:
[0046] In the present invention, by using time series analysis to conduct in-depth analysis of current and voltage waveforms, it is possible to identify subtle offsets and trend changes, adopt dynamic threshold configuration, and combine historical and real-time data for comprehensive analysis through Fourier transform, allowing automatic adjustment of thresholds under various working conditions, reducing misoperation and improving the accuracy of protection response. Driven by continuously optimized data, the protection mechanism can self-adjust to adapt to new or changing grid conditions, greatly reducing the risks caused by equipment failure or operating errors, and ensuring the continuous stable operation and safety of the power system. BRIEF DESCRIPTION OF THE DRAWINGS
[0047] Figure 1 It is a schematic diagram of the workflow of the present invention;
[0048] Figure 2 This is a detailed flow chart of S1 of the present invention;
[0049] Figure 3 This is a detailed flow chart of S2 of the present invention;
[0050] Figure 4 This is a detailed flow chart of S3 of the present invention;
[0051] Figure 5 This is a detailed flow chart of S4 of the present invention;
[0052] Figure 6 This is a detailed flow chart of S5 of the present invention;
[0053] Figure 7 This is a detailed flow chart of S6 of the present invention;
[0054] Figure 8 It is a system flow chart of the present invention. DETAILED DESCRIPTION
[0055] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.
[0056] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.
[0057] Example 1: Please refer to Figure 1 The present invention provides a technical solution: a relay circuit short circuit protection method, comprising the following steps:
[0058] S1: Based on the output data of the current sensor, separate the mutation points and peak values in the current and voltage data, establish the association model between the morphological characteristics and the circuit state, formulate the data classification standard, and generate the circuit state feature set;
[0059] S2: Based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of the current and voltage waveforms, extract the change trend characteristics, and identify and classify the feature patterns to generate an abnormal waveform feature set;
[0060] S3: Based on the abnormal waveform feature set, the cross-validation technology is used to adjust the model weights and deviation values through current data feedback, and the stability and accuracy of the model are verified by repeated experiments to generate a short circuit prediction model;
[0061] S4: Based on the short-circuit prediction model, percentiles are used in combination with historical waveform data to calculate the threshold for responding to sudden current increases through Fourier transform, and the threshold is dynamically adjusted and the response mechanism is set according to the overcurrent protection requirements to generate a dynamic threshold configuration;
[0062] S5: Based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism, records the status of the protection action, and generates circuit breaker protection status information;
[0063] S6: Based on the circuit breaker protection status information, collect the relay circuit performance data after the protection action is executed, evaluate the response effect, and refine the short-circuit prediction model and threshold to generate the protection mechanism optimization result.
[0064] The circuit status feature set includes voltage extremes, current peaks, and sampling time intervals. The abnormal waveform feature set includes waveform deviation, change trends, and classification indicators. The short circuit prediction model includes parameter optimization records and verification accuracy results. The dynamic threshold configuration includes real-time adjustment logic, historical threshold data, and emergency response levels. The circuit breaker protection status includes activation delay, trigger threshold, and event log. The protection mechanism optimization results include performance improvement indicators, adjustment plans, and efficiency evaluation results.
[0065] See also Figure 2 Based on the output data of the current sensor, the mutation points and peak values in the current and voltage data are separated, the association model between the morphological characteristics and the circuit state is established, and the data classification standard is formulated. The specific steps of generating the circuit state feature set are as follows:
[0066] S101: Based on the output data of the current sensor, by real-time monitoring of current and voltage readings, marking the mutation points and peak values of the data, and recording the timestamp and amplitude of each mutation point, to obtain a mutation point and peak value data set;
[0067] S101: Based on the output data of the current sensor, the real-time data of the current and voltage in the circuit are continuously collected by measuring equipment. The collection time interval of each data point is set to 1 millisecond to ensure high density and high accuracy of the data. Quantitative analysis technology is used to identify mutation points in the data, including marking mutations by identifying rapid changes in current or voltage. For each marked mutation point, the exact timestamp and change amplitude are recorded. This is automatically completed by the data processor, and the records form a detailed mutation point and peak data set.
[0068] S102: Based on the mutation point and peak data set, analyzing the occurrence pattern of each mutation point and peak in the data set, building a corresponding relationship between the data point and the circuit working state, and obtaining a primary circuit state model by comparing the data patterns under the difference state;
[0069] S102: Based on the mutation point and peak data set, the occurrence frequency and pattern of each mutation point and peak in the data set are analyzed by statistical methods, and data mining techniques such as cluster analysis are used to classify the mutation points and peaks, and identify the typical data patterns of the circuit under different working conditions. In this way, the real-time working state of the circuit can be matched with the specific data pattern, and a primary circuit state model can be constructed. This model not only reflects the behavior of the circuit under different states such as normal, overload and short circuit, but also can further refine the judgment logic of the circuit state by comparing the differences between the states.
[0070] S103: Based on the primary circuit state model, perform data feature classification processing, sort and encode the data according to the circuit state, and integrate the classified data to obtain a circuit state feature set;
[0071] S103: Based on the primary circuit state model, the data features are then classified and processed in detail. The data points are sorted and encoded according to the circuit state defined by the model. Each state is assigned a unique code. Subsequently, data integration techniques such as vector quantization are used to aggregate the encoded data into a comprehensive circuit state feature set. This set is not only a simple summary of the data, but through data analysis and pattern recognition technology, the data is converted into a powerful tool for circuit performance and state evaluation. Through such processing, the circuit state can be intuitively expressed and utilized.
[0072] See also Figure 3 ,Based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of current and voltage waveforms, extract the change trend characteristics, and identify and classify the characteristic patterns. The specific steps for generating the abnormal waveform feature set are as follows:
[0073] S201: Based on the circuit state feature set, using a recurrent neural network, analyzing the data waveforms of current and voltage, marking the offset point of each data waveform, and recording the amplitude and frequency of the offset point, analyzing the offset rule, and obtaining a waveform offset data set;
[0074] S201: Based on the circuit state feature set, a recurrent neural network model is set up to process time series data. The current and voltage data of each acquisition cycle are deeply analyzed. The feature extraction capability of the neural network is used to identify and mark the offset points in the data waveform. The offset points refer to the points where the waveform obviously deviates from the normal mode. Specifically, by analyzing the differences between consecutive data points, the amplitude and frequency of the offset points are automatically recorded. The data is used to form a detailed waveform offset data set. Through the comprehensive analysis of the offset points, the regularity and common patterns of the offsets are identified, thereby providing a more accurate data basis for circuit state analysis.
[0075] Recurrent neural network, according to the formula: Calculate the hidden state at each time step, identify and mark the offset points, and generate the amplitude and frequency data of the offset points, where is the time step The hidden state of is the time step The hidden state of is the time step The input data, is the weight matrix from hidden state to hidden state, is the weight matrix input to the hidden state, is the deviation term, is the activation function, , and are the weight adjustment coefficients for hidden state, input data, and context state, respectively. is the time step The context state, is the weight matrix from context state to hidden state.
[0076] Execution process:
[0077] First, calculate the time step The hidden state of and contextual state The product of And adjust the coefficient , used to control the impact of historical information on the current state, then the time step Input data Multiply by weight and apply the adjustment factor , to optimize the contribution of the input data to the hidden state, and then and Multiply and use coefficients Adjust to incorporate contextual information, and finally, add the bias to the result , and through the activation function Transformation, generating new hidden states , which is used for prediction of the next time step and identification of the offset point, and the weight coefficient , and The determination of should be automatically adjusted through gradient descent on the training dataset to optimize the model performance.
[0078] S202: extracting the rising or falling trend of each waveform based on the waveform offset data set, classifying the change trend, and recording the waveform features in each classification to obtain a primary trend feature set;
[0079] S202: Based on the waveform offset data set, further analyze the specific behavior pattern of each waveform, especially its rising or falling trend, through data analysis tools to identify and record the changing trend of the waveform under different conditions. Subsequently, the changing trend is classified into different categories, and each category is divided according to the specific characteristics of the waveform. For example, fast-rising waveforms and slowly falling waveforms are divided into different categories. The classification not only helps to understand the background mechanism of the waveform behavior, but also provides a basis for further trend prediction. In this way, a primary trend feature set containing the rising and falling trends of the waveform is constructed.
[0080] S203: Based on the primary trend feature set, iteratively refine the features, group the waveform features according to functionality or abnormality, organize the groups into standard classifications, and obtain an abnormal waveform feature set;
[0081] S203: Based on the primary trend feature set, the next step is to iteratively refine the features in more depth. By analyzing and comparing different waveform features, they are grouped clearly according to functionality or abnormality. For example, waveforms that exhibit similar abnormal behaviors are grouped together for special processing and monitoring. This grouping not only makes the data more orderly, but also facilitates subsequent data monitoring and abnormal warning. Through such processing, a systematic abnormal waveform feature set is compiled, which records the characteristics and classification of various abnormal waveforms.
[0082] See also Figure 4 , based on the abnormal waveform feature set, using cross-validation technology, adjusting the model weights and bias values through current data feedback, and repeating the test to verify the stability and accuracy of the model, the steps to generate the short circuit prediction model are as follows:
[0083] S301: Based on the abnormal waveform feature set, multiple sets of verification data sets are set to perform model testing, the impact of weight adjustment on the test results is analyzed, the model performance of each set of data is recorded, and multiple rounds of model performance verification are performed to generate a preliminary model data set;
[0084] S301: Based on the abnormal waveform feature set, multiple sets of verification data sets with different characteristics are configured to test and evaluate the performance of the model. The verification data sets are screened from real circuit data, covering current and voltage waveforms from normal to various abnormal states. In the model testing phase, special attention is paid to the adjustment of weights and their impact on the model test results. By recording the performance of each set of data in the model, including key performance indicators such as prediction accuracy and response time, each weight configuration is verified multiple times. Through this iterative testing method, the optimal weight setting can be gradually adjusted and determined to generate a preliminary model data set containing preliminary test data and weight configuration.
[0085] S302: Based on the preliminary model data set, analyze the stability and accuracy of multiple test results, identify the best performing weight and bias value configuration, adjust the model parameters, optimize the accuracy and response speed of the model, and obtain an optimized model data set;
[0086] S302: Based on the preliminary model data set, further evaluate the performance of the model, especially the stability and accuracy of the test results. By analyzing the results of multiple consecutive tests, identify the key factors affecting the model performance, such as the optimal configuration of weights and bias values, and accurately adjust the parameters to improve the accuracy and response speed of the model. This optimization process involves data analysis technology and experimental design to ensure that the model can maintain high performance under different test conditions. Through the steps, an optimized model data set is obtained.
[0087] S303: Based on the optimized model data set, repeatedly perform multiple rounds of tests to evaluate the performance of the model under various conditions, determine the performance consistency of the model on the difference test set, and generate a short circuit prediction model;
[0088] S303: Based on the optimized model data set, continue to perform multiple rounds of testing. This process not only tests the accuracy of the model, but also focuses on evaluating its performance consistency under different conditions. By repeating the test on multiple test sets, the model's ability to predict various types of abnormal current and voltage waveforms can be evaluated. After each round of testing, the model's performance data is collected and analyzed, including prediction accuracy, stability, and sensitivity to abnormal conditions. The data helps determine the reliability and effectiveness of the model in practical applications and generates a short-circuit prediction model that can effectively predict short-circuit situations in circuits.
[0089] See also Figure 5 ,According to the short-circuit prediction model, percentiles are used in combination with historical waveform data, and the threshold for responding to sudden current increases is calculated through Fourier transform. According to the overcurrent protection requirements, the threshold is dynamically adjusted and the response mechanism is set. The steps for generating dynamic threshold configuration are as follows:
[0090] S401: Based on the short circuit prediction model and in combination with historical waveform data, a preliminary threshold is set by calculating the percentile of the historical data, and the performance of the threshold under the difference waveform is analyzed to generate an initial threshold set;
[0091] S401: Based on the short-circuit prediction model, historical current and voltage waveform data are integrated and analyzed. On this basis, preliminary thresholds are set by calculating the percentiles of the current and voltage waveforms in the historical data. The percentiles are determined based on the distribution characteristics of the current and voltage in the historical waveform data. For example, the 90% percentile is used as the threshold to prevent sudden high peak values. Next, the performance of the preliminary threshold under different current and voltage waveforms is analyzed, especially the effects under abnormal and unconventional waveform conditions. In this way, an initial threshold set that reflects various circuit conditions can be generated.
[0092] S402: Based on the initial threshold set, dynamically adjust the threshold according to the current increase data monitored in real time through Fourier transform, record the parameters and results of each adjustment, analyze the impact on the current protection effect, optimize the threshold adjustment process, and obtain the adjusted threshold set;
[0093] S402: Based on the initial threshold set, the current gain data monitored in real time is further analyzed using the Fourier transform method. This transformation can help identify the periodic and non-periodic change characteristics in the current data, so as to adjust the threshold more accurately. Each time the threshold is dynamically adjusted based on real-time data, the specific parameters and results of the adjustment are recorded, including the current protection effect before and after the threshold adjustment. Through this method, the threshold adjustment process is continuously analyzed and optimized to ensure that the threshold can effectively respond to the actual changes in the circuit, forming a carefully adjusted threshold set, which includes thresholds adjusted for various circuit states.
[0094] S403: Setting a current surge response mechanism based on the adjusted threshold set, testing the real-time response effect of the threshold by simulating a differential current surge situation, verifying the effectiveness of the threshold setting and the response mechanism, and generating a dynamic threshold configuration;
[0095] S403: Based on the adjusted threshold set, a current surge response mechanism is set. The mechanism mainly simulates different current surge situations to test the response effect of the threshold in actual applications. Through simulation tests, the rationality of the threshold setting and the effectiveness of the response mechanism can be verified. During the test, the response time and accuracy of the threshold to the sudden increase in current are observed and recorded, and its performance under various current change scenarios is evaluated. The test results help to further adjust the threshold configuration to ensure its dynamic and effective response in actual circuit operation, and generate a dynamic threshold configuration, which not only provides an optimized setting of the threshold, but also ensures the safety and stability of the circuit when facing a current surge.
[0096] See also Figure 6 , based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism and records the status of the protection action. The specific steps for generating circuit breaker protection status information are:
[0097] S501: Based on the dynamic threshold configuration, the newly received current and voltage data are continuously monitored. If the detection data exceeds the set threshold, the protection mechanism of the circuit breaker is activated, and the triggering time and environmental parameters of the protection action are recorded to generate a protection event record;
[0098] S501: Based on the dynamic threshold configuration, the monitoring system is implemented to continuously track the current and voltage data in the circuit. Through the real-time data processing unit, each current and voltage reading is detected in real time. Once the data exceeds the preset dynamic threshold, the protection mechanism of the circuit breaker is immediately activated. The protection action includes disconnecting the power supply to prevent circuit damage. At the same time, the time when the protection action is triggered and the relevant environmental parameters, such as temperature, humidity and circuit load conditions, are accurately recorded. The detailed records are automatically integrated into the protection event record by the system.
[0099] S502: Based on the protection event records, analyze the time point, environmental conditions and thresholds of each circuit breaker triggering, evaluate the reaction speed and triggering accuracy of the protection mechanism, and generate protection effect analysis results;
[0100] S502: Based on the protection event records, analyze the specific time point, environmental conditions and threshold settings of each circuit breaker triggering. Through analysis, the reaction speed and triggering accuracy of the circuit breaker protection mechanism can be evaluated. During the analysis, special attention is paid to whether the protection mechanism can respond in time before dangerous changes occur in the circuit, and whether the protection action is always triggered at the appropriate threshold. The generated protection effect analysis results will directly affect the adjustment and optimization of future protection strategies.
[0101] S503: Based on the protection effect analysis result, data of difference protection events are summarized and compared, the frequency, state and environmental response of the events are recorded, and circuit breaker protection state information is generated;
[0102] S503: Based on the protection effect analysis results, further summarize and compare the data differences between different protection events, and form a comprehensive circuit breaker protection status information report by statistically analyzing the frequency of protection events, responses under different conditions, and the influence of environmental factors. The report not only summarizes the processing effects of various protection events, but also reveals the performance of the protection mechanism under different conditions, and provides an important basis for improving protection strategies and optimizing threshold configurations. In this way, the safe operation of the circuit system is ensured, and the efficiency and reliability of the entire protection mechanism are improved.
[0103] See also Figure 7 , based on the circuit breaker protection status information, collect the relay circuit performance data after the protection action is executed, evaluate the response effect, and refine the short-circuit prediction model and threshold to generate the protection mechanism optimization results. The specific steps are:
[0104] S601: Based on the circuit breaker protection status information, collect the relay circuit performance data after the protection action, analyze the performance of the relay circuit under different working conditions, evaluate the stability and response efficiency, and generate the relay performance analysis results;
[0105] S601: Based on the circuit breaker protection status information, continue to conduct subsequent analysis on the circuit breaker protection mechanism and collect the relay circuit performance data after different protection actions are triggered. The data includes key performance indicators such as current, voltage and response time of the relay circuit under different working conditions. Through the data, a comprehensive analysis is conducted to evaluate the stability and response efficiency of the relay circuit under various environments and working conditions. The process involves data induction, comparison and trend analysis to generate relay performance analysis results. The results reveal the actual performance of the relay circuit and potential improvement directions.
[0106] S602: Based on the relay performance analysis results, analyze the impact of the protection action on the relay circuit performance, adjust the parameters of the short circuit prediction model and threshold, and continuously monitor and test the effectiveness of the adjustment to generate the protection mechanism adjustment configuration;
[0107] S602: Based on the relay performance analysis results, further analyze the specific impact of the circuit breaker protection action on the relay circuit performance, pay special attention to the recovery of the relay circuit after the circuit breaker is triggered and the long-term performance stability. According to the analysis results, adjust the parameters of the short-circuit prediction model and dynamic threshold to improve the prediction accuracy and adaptability of the threshold. The adjustment is based on actual data feedback and simulation test results to ensure that the adjustment measures can effectively respond to various circuit conditions. After the adjustment, implement continuous monitoring and multiple rounds of test verification to evaluate the effectiveness of the adjustment. Through continuous testing and monitoring, generate a detailed report containing the protection mechanism adjustment configuration.
[0108] S603: Based on the protection mechanism adjustment configuration, the short circuit prediction model and threshold settings are adjusted, multiple rounds of testing are performed to verify the adaptability and accuracy of the adjusted model and threshold under various conditions, and generate protection mechanism optimization results;
[0109] S603: Based on the protection mechanism adjustment configuration, further optimization and adjustment are made to the short circuit prediction model and threshold settings, including multiple rounds of rigorous testing to verify the adaptability and accuracy of the adjusted model and threshold under different test conditions. The test not only verifies the performance of the model and threshold under standard conditions, but also evaluates the response capabilities under extreme and uncommon conditions. After each round of testing, data is collected and analyzed to ensure that each adjustment improves the overall effect of the protection mechanism. Through this comprehensive optimization process, the optimization results of the protection mechanism are generated, which provide higher reliability and efficiency for circuit protection.
[0110] See also Figure 8, a relay circuit short circuit protection system, the system comprising:
[0111] The current and voltage data analysis module samples the current and voltage data based on the output data of the current sensor, separates the mutation point and the peak value, and establishes the corresponding relationship with the circuit state by analyzing the morphological characteristics of the current and voltage to generate a circuit state feature set;
[0112] The time series analysis module performs time series analysis on the current and voltage waveforms based on the circuit state feature set, identifies offsets, extracts trend features of waveform changes, identifies and classifies features, and generates an abnormal waveform feature set;
[0113] The model optimization and verification module performs model parameter optimization based on the abnormal waveform feature set, uses cross-validation technology to adjust model weights and deviation values, verifies the stability and accuracy of the model through data feedback and repeated tests, and generates a short circuit prediction model;
[0114] The threshold dynamic configuration module calculates the threshold of the sudden current based on the analysis results of the short-circuit prediction model, using percentiles and historical waveform data through Fourier transform, dynamically adjusts the threshold and sets the response mechanism according to the protection requirements, and generates dynamic threshold configuration;
[0115] The real-time monitoring response module continuously analyzes the newly received current and voltage data based on dynamic threshold configuration. When the data exceeds the set threshold, the circuit breaker protection mechanism is triggered, and the performance data of the relay circuit after protection is collected to evaluate the effect. The short-circuit prediction model and threshold configuration are refined and adjusted to generate protection mechanism optimization results.
[0116] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.
Claims
1. A relay circuit short circuit protection method, characterized in that: The following steps are involved: Based on the output data of the current sensor, the mutation points and peak values in the current and voltage data are separated, a correlation model between the morphological features and the circuit state is established, and a data classification standard is formulated to generate a circuit state feature set; Based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of current and voltage waveforms, extract change trend characteristics, and identify and classify feature patterns to generate an abnormal waveform feature set; Based on the abnormal waveform feature set, cross-validation technology is used to adjust the model weights and deviation values through current data feedback, and repeated experiments are performed to verify the stability and accuracy of the model to generate a short circuit prediction model; According to the short circuit prediction model, using percentiles, combined with historical waveform data, and through Fourier transform, the threshold for responding to sudden current increase is calculated, and according to the overcurrent protection requirements, the threshold is dynamically adjusted and the response mechanism is set to generate a dynamic threshold configuration; Based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism, records the status of the protection action, and generates circuit breaker protection status information; Based on the circuit breaker protection status information, the relay circuit performance data after the protection action is executed is collected to evaluate the response effect, and the short circuit prediction model and threshold are refined and adjusted to generate a protection mechanism optimization result; Based on the circuit state feature set, a recurrent neural network is used to perform time series analysis, identify the offset of current and voltage waveforms, extract change trend characteristics, and identify and classify feature patterns. The specific steps of generating an abnormal waveform feature set are as follows: Based on the circuit state feature set, a recurrent neural network is used to analyze the data waveforms of current and voltage, mark the offset point of each data waveform, and record the amplitude and frequency of the offset point, analyze the offset rule, and obtain a waveform offset data set; Based on the waveform offset data set, extract the rising or falling trend of each waveform, classify the change trend, and record the waveform features in each classification to obtain a primary trend feature set; Based on the primary trend feature set, the features are iteratively refined, the waveform features are grouped according to functionality or abnormality, and the groups are sorted into standard classifications to obtain an abnormal waveform feature set.
2. The relay circuit short circuit protection method according to claim 1, characterized in that: The circuit state feature set includes voltage extremes, current peaks, and sampling time intervals; the abnormal waveform feature set includes waveform deviation, change trends, and classification indicators; the short circuit prediction model includes parameter optimization records and verification accuracy results; the dynamic threshold configuration includes real-time adjustment logic, historical threshold data, and emergency response levels; the circuit breaker protection state includes activation delay, trigger threshold, and event log; the protection mechanism optimization results include performance improvement indicators, adjustment plans, and efficiency evaluation results.
3. The relay circuit short circuit protection method according to claim 1, characterized in that: Based on the output data of the current sensor, the mutation points and peak values in the current and voltage data are separated, the association model between the morphological features and the circuit state is established, and the data classification standard is formulated. The specific steps for generating the circuit state feature set are as follows: Based on the output data of the current sensor, the current and voltage readings are monitored in real time, the mutation points and peak values of the data are marked, and the timestamp and amplitude of each mutation point are recorded to obtain the mutation point and peak value data set; Based on the mutation point and peak value data set, analyzing the occurrence pattern of each mutation point and peak value in the data set, constructing the corresponding relationship between the data point and the circuit working state, and obtaining the primary circuit state model by comparing the data patterns under the difference state; Based on the primary circuit state model, data feature classification processing is performed, the data is sorted and encoded according to the circuit state, and the classified data is integrated to obtain a circuit state feature set.
4. The relay circuit short circuit protection method according to claim 1, characterized in that: The recurrent neural network, according to the formula: Calculate the hidden state at each time step, identify and mark the offset points, and generate the amplitude and frequency data of the offset points, where is the time step The hidden state of is the time step The hidden state of is the time step The input data, is the weight matrix from hidden state to hidden state, is the weight matrix input to the hidden state, is the deviation term, is the activation function, , and are the weight adjustment coefficients for hidden state, input data, and context state, respectively. is the time step The context state, is the weight matrix from context state to hidden state.
5. The relay circuit short circuit protection method according to claim 1, characterized in that: Based on the abnormal waveform feature set, cross-validation technology is used to adjust the model weights and deviation values through current data feedback, and repeated experiments are performed to verify the stability and accuracy of the model. The specific steps for generating a short circuit prediction model are as follows: According to the abnormal waveform feature set, multiple sets of validation data sets are set to perform model testing, the impact of weight adjustment on the test results is analyzed, the model performance of each set of data is recorded, and multiple rounds of model performance verification are performed to generate a preliminary model data set; Based on the preliminary model data set, analyzing the stability and accuracy of multiple test results, identifying the best performing weight and bias value configuration, adjusting the model parameters, optimizing the accuracy and response speed of the model, and obtaining an optimized model data set; Based on the optimized model data set, multiple rounds of testing are repeatedly performed to evaluate the performance of the model under various conditions, determine the performance consistency of the model on the difference test set, and generate a short circuit prediction model.
6. The relay circuit short circuit protection method according to claim 1, characterized in that: According to the short circuit prediction model, percentiles are used in combination with historical waveform data, and the threshold for responding to sudden current increases is calculated through Fourier transform. The threshold is dynamically adjusted and the response mechanism is set according to the overcurrent protection requirements. The steps for generating dynamic threshold configuration are as follows: Based on the short circuit prediction model, combined with historical waveform data, the initial threshold is set by calculating the percentile of the historical data, and the performance of the threshold under the difference waveform is analyzed to generate an initial threshold set; Based on the initial threshold set, dynamically adjust the threshold according to the current increase data monitored in real time through Fourier transform, record the parameters and results of each adjustment, analyze the impact on the current protection effect, optimize the threshold adjustment process, and obtain the adjusted threshold set; Based on the adjusted threshold set, a current surge response mechanism is set, and by simulating a differential current surge situation, the real-time response effect of the threshold is tested, the effectiveness of the threshold setting and the response mechanism is verified, and a dynamic threshold configuration is generated.
7. The relay circuit short circuit protection method according to claim 1, characterized in that: Based on the dynamic threshold configuration, the newly received current and voltage data are analyzed in real time. When the real-time monitoring data exceeds the threshold, the circuit breaker automatically triggers the protection mechanism and records the status of the protection action. The specific steps of generating the circuit breaker protection status information are as follows: Based on the dynamic threshold configuration, the newly received current and voltage data are continuously monitored. If the detected data exceeds the set threshold, the protection mechanism of the circuit breaker is activated, and the triggering time and environmental parameters of the protection action are recorded to generate a protection event record; Based on the protection event records, analyze the time point, environmental conditions and thresholds of each circuit breaker triggering, evaluate the reaction speed and triggering accuracy of the protection mechanism, and generate protection effect analysis results; Based on the protection effect analysis results, data of difference protection events are summarized and compared, the frequency, status and environmental response of the events are recorded, and circuit breaker protection status information is generated.
8. The relay circuit short circuit protection method according to claim 1, characterized in that: Based on the circuit breaker protection status information, the relay circuit performance data after the protection action is executed is collected, the response effect is evaluated, and the short circuit prediction model and threshold are refined and adjusted to generate the protection mechanism optimization result. The specific steps are: Based on the circuit breaker protection status information, collect the relay circuit performance data after the protection action, analyze the performance of the relay circuit under different working conditions, evaluate the stability and response efficiency, and generate the relay performance analysis results; Based on the relay performance analysis results, the impact of the protection action on the relay circuit performance is analyzed, the parameters of the short circuit prediction model and the threshold are adjusted, and the effectiveness of the adjustment is continuously monitored and tested to verify, and a protection mechanism adjustment configuration is generated; Based on the protection mechanism adjustment configuration, the short circuit prediction model and threshold settings are adjusted, multiple rounds of testing are performed to verify the adaptability and accuracy of the adjusted model and thresholds under various conditions, and generate protection mechanism optimization results.
9. A relay circuit short circuit protection system, characterized in that: According to the relay circuit short circuit protection method according to any one of claims 1 to 8, the system comprises: The current and voltage data analysis module samples the current and voltage data based on the output data of the current sensor, separates the mutation point and the peak value, and establishes the corresponding relationship with the circuit state by analyzing the morphological characteristics of the current and voltage to generate a circuit state feature set; The time series analysis module performs time series analysis on the current and voltage waveforms based on the circuit state feature set, performs offset identification, extracts trend features of waveform changes, identifies and classifies the features, and generates an abnormal waveform feature set; The model optimization and verification module performs model parameter optimization based on the abnormal waveform feature set, uses cross-validation technology to adjust model weights and deviation values, verifies the stability and accuracy of the model through data feedback and repeated tests, and generates a short circuit prediction model; The threshold dynamic configuration module calculates the threshold of the burst current by using percentiles and combining historical waveform data according to the analysis results of the short circuit prediction model, dynamically adjusts the threshold and sets the response mechanism according to the protection requirements, and generates a dynamic threshold configuration; The real-time monitoring response module continuously analyzes the newly received current and voltage data based on the dynamic threshold configuration. When the data exceeds the set threshold, the circuit breaker protection mechanism is triggered, and the performance data of the relay circuit after protection is collected to evaluate the effect, refine and adjust the short circuit prediction model and threshold configuration, and generate the protection mechanism optimization result.
