A bypass shelter circuit switching-based electric leakage protection method and system

Abstract

The application discloses a bypass shelter circuit switching-based electric leakage protection method and system, which comprises the following steps: continuously collecting voltage and current data of multiple nodes in a bypass shelter circuit switching process in real time, and forming a multi-point electric leakage initial signal set during load transfer; performing dynamic mode recognition by using a convolutional neural network, extracting a feature vector for risk judgment, and identifying potential electric leakage risk nodes; acquiring auxiliary channel configuration parameters, calculating and determining one or more target auxiliary channels for shunting electric leakage current in real time, and generating corresponding channel switching instructions; activating the target auxiliary channel to establish an electric leakage current shunting path; dynamically adjusting the impedance of the target auxiliary channel to make the distribution of shunted current meet the load transfer stability requirement, and completing circuit switching without interruption. The application can realize intelligent early warning and active diversion of dynamic multi-point electric leakage risk, and realizes electric leakage protection without power interruption in the bypass shelter circuit switching process.

Description

Technical Field

[0001] This invention relates to the field of bypass cabin circuit switching technology, and in particular to a leakage current protection method and system based on bypass cabin circuit switching. Background Technology

[0002] In power supply systems, circuit switching is a core operation to ensure power continuity and safe equipment operation. This requirement is particularly prominent in temporary or emergency power supply scenarios such as bypass shelters. The core objective of such operations is to safely transfer the load from one line to another without interrupting power. Any interruption or error may lead to power instability or even equipment safety accidents. Among the many risks, leakage current, due to its hidden and dynamic nature, remains the primary factor threatening the safety of operators and the integrity of equipment. Leakage current can not only directly damage precision equipment and cause electric shock accidents, but in severe cases, it can also cause electric arc fires, leading to the paralysis of the entire power supply system.

[0003] For a long time, the industry has developed some conventional solutions for leakage current protection. The most common approach is to rely on strengthening the insulation level of lines and equipment, or installing leakage current devices (RCDs). The former is a passive defense, and its protective capability will significantly decrease when equipment ages or the on-site environment is complex (such as humidity and high temperature). The latter is a reactive protection, that is, to quickly cut off the circuit after a leakage current is detected. However, in the dynamic process of circuit switching in bypass cabins, both methods have exposed fundamental defects. The biggest problem is that traditional methods cannot cope with the risk of dynamic and changing multi-point leakage current. During the switching process, the load and circuit status change rapidly. Once a leakage current occurs, existing technology cannot quickly and accurately locate the fault point in complex temporary circuits, resulting in excessively long troubleshooting time. This not only delays the restoration of power supply, but may also trigger cascading failures under high loads. More importantly, simple power outage protection contradicts the fundamental requirement of uninterrupted power switching, and cannot maintain the continuity of power supply while ensuring safety. Summary of the Invention

[0004] Therefore, the technical problem to be solved by the present invention is to overcome the defects of the prior art in which dynamic leakage risk cannot be monitored in real time when the bypass cabin circuit is switched and the power supply continuity is affected by the reliance on power outage protection. The present invention provides a leakage protection method based on bypass cabin circuit switching, which can realize intelligent early warning and active diversion of dynamic multi-point leakage risk, thereby significantly improving operational safety and system reliability without interrupting power supply.

[0005] To address the aforementioned technical problems, this invention provides a leakage current protection method based on bypass cabin circuit switching, comprising the following steps:

[0006] Real-time continuous acquisition of voltage and current data of multiple nodes during the switching process of the bypass cabin circuit constitutes a set of initial leakage signals at multiple points during load transfer.

[0007] A convolutional neural network is used to perform dynamic pattern recognition on the initial signal set of multiple leakage currents, extract feature vectors, and make risk judgments based on the feature vectors to identify potential leakage current risk nodes.

[0008] The auxiliary channel configuration parameters are obtained from the identified potential leakage risk nodes, one or more target auxiliary channels for shunt leakage current are calculated and determined in real time, and corresponding channel switching instructions are generated.

[0009] In response to a channel switching command, the target auxiliary channel is activated to establish a leakage current shunt path;

[0010] The impedance of the target auxiliary channel is dynamically adjusted to ensure that the distribution of the shunt current meets the load transfer stability requirements, thus completing uninterrupted circuit switching.

[0011] In one embodiment of the present invention, voltage and current data of multiple nodes are continuously collected in real time during the switching process of the bypass cabin circuit to form a multi-point leakage current initial signal set during load transfer, specifically including:

[0012] Multiple voltage-current sensing units are deployed at the bus interface, load access point, and key switching branch nodes of the bypass cabin circuit to form a distributed sensing network covering the circuit architecture.

[0013] Broadcast a unified synchronization acquisition command to the distributed sensor network to coordinate all voltage-current sensing units to start data acquisition with the same sampling clock;

[0014] Inside each voltage-current sensing unit, the acquired raw analog signal is differentially amplified and hardware filtered to obtain the denoised local signal.

[0015] The local signals of each sensing unit after denoising are converted from analog to digital and stamped with a unified timestamp and node identifier to form a standardized node data packet.

[0016] All node data packets are aligned and reassembled based on timestamps and node identifiers to generate a multi-point leakage initial signal set.

[0017] In one embodiment of the present invention, a convolutional neural network is used to perform dynamic pattern recognition on a set of initial signals of multi-point leakage current and extract feature vectors, specifically including:

[0018] The initial signals of multi-point leakage current are standardized and segmented in the time dimension to construct a two-dimensional signal spectrum with time as the horizontal axis and multi-node signals as the vertical axis.

[0019] Construct a feature extraction network with a hierarchical perception structure. The front-end structure of this network is used to perform local feature scanning on the two-dimensional signal spectrum and capture the local correlation patterns of the signal in the time and node dimensions.

[0020] By utilizing the deep structure of the feature extraction network, multi-level fusion and abstraction are performed on the local correlation patterns obtained from scanning to form a high-order temporal feature combination;

[0021] The high-order temporal features are combined and mapped into a fixed-dimensional comprehensive feature vector that condenses the dynamic switching state information of the current time period.

[0022] In one embodiment of the present invention, risk determination is performed based on feature vectors to identify potential leakage risk nodes, specifically including:

[0023] The comprehensive feature vector is input into a pre-trained state assessment model, which is trained based on feature vector data from historical normal operation and known leakage states;

[0024] The status assessment model outputs a three-dimensional risk assessment result, which includes a risk level score, a risk type identifier, and a list of nodes with the highest risk contribution.

[0025] Based on the preset dynamic risk strategy, the risk assessment results are analyzed. If the risk level score exceeds the dynamic warning line under the current system load, risk confirmation is triggered.

[0026] After risk confirmation is triggered, one or more potential leakage risk nodes are identified by combining the risk type identifier with the list of nodes with the highest risk contribution.

[0027] In one embodiment of the present invention, auxiliary channel configuration parameters are obtained from identified potential leakage risk nodes, and one or more target auxiliary channels for shunt leakage current are calculated and determined in real time, specifically including:

[0028] Based on the location information of potential leakage risk nodes, all electrically connected backup auxiliary channels are retrieved from the preset channel topology to form an initial candidate channel set;

[0029] Based on the estimated leakage intensity of potential leakage risk nodes and the real-time load capacity of each candidate channel, current carrying capacity matching calculation is performed to screen out all channels with qualified load-bearing capacity, forming a qualified channel subset.

[0030] For each channel in the qualified channel subset, simulate and calculate its impact on the power flow distribution of the entire circuit system after it is put into operation, and eliminate channels that would cause any node voltage to exceed the limit or equipment to overload, thus forming a safe channel subset;

[0031] From the subset of safe channels, select the channel with the shortest path and the smallest equivalent impedance, and determine it as the target auxiliary channel.

[0032] In one embodiment of the present invention, when there is no single channel in the subset of safe channels that meets the current shunting requirements, one or more target auxiliary channels for shunting leakage current are calculated and determined in real time, further including:

[0033] Based on the qualified channel subset, generate all possible parallel cooperative channel combinations consisting of two or more channels;

[0034] For each parallel and coordinated channel combination, a coordinated current splitting model is constructed to calculate the overall power quality index and thermal stability margin of the system when the leakage current is distributed among the channels in a specific ratio.

[0035] From all feasible combinations of collaborative channels, the combination that yields the best overall power quality index and the largest thermal stability margin is selected as the optimal collaborative diversion scheme.

[0036] Based on the optimal collaborative diversion scheme, a collaborative switching instruction set is generated, which includes the deployment timing of each target auxiliary channel and its preset diversion weight, as the final channel switching instruction.

[0037] In one embodiment of the present invention, the step of determining the target auxiliary channel from a subset of secure channels further includes performing system-level conflict verification:

[0038] Obtain the total power of all currently running loads in the system, as well as the remaining capacity of the upstream power supply for each candidate auxiliary channel;

[0039] The estimated leakage current intensity of potential leakage risk nodes is superimposed with the additional shunt power required by the candidate channel to calculate the total power demand after the channel is put into operation.

[0040] Compare the total power demand with the remaining capacity of the corresponding upstream power supply to verify whether the power supply capacity is sufficient.

[0041] The target auxiliary channel is finally determined from the candidate channels that pass the verification.

[0042] In one embodiment of the present invention, dynamically adjusting the impedance of the target auxiliary channel specifically includes:

[0043] Real-time monitoring of the actual shunt current value flowing through the target auxiliary channel;

[0044] The actual shunt current value is compared with the target current range calculated based on the load transfer stability requirements to generate a current deviation signal;

[0045] Based on the magnitude and direction of the current deviation signal, a corresponding impedance adjustment command is generated; if the actual value is higher than the upper limit of the target range, the command is to increase the impedance; if the actual value is lower than the lower limit of the target range, the command is to decrease the impedance.

[0046] In one embodiment of the present invention, generating a corresponding impedance adjustment command based on the magnitude and direction of the current deviation signal further includes:

[0047] Analyze the changing trend of the current deviation signal to predict its next direction and magnitude of change;

[0048] Based on the current actual shunt current value and its prediction information, a comprehensive combination of adjustment parameters is retrieved from a preset adjustment strategy mapping table. This combination defines the intensity and rate of this adjustment.

[0049] Based on the combination of adjustment parameters, a smooth, phased sequence of impedance adjustment commands is generated to replace a single instantaneous adjustment command.

[0050] To address the aforementioned technical problems, this invention also provides a leakage current protection system based on bypass cabin circuit switching, used to implement the above-mentioned leakage current protection method, comprising:

[0051] The data acquisition module is used to continuously acquire voltage and current data of multiple nodes during the switching process of the bypass cabin circuit in real time, forming a set of initial leakage signals at multiple points during load transfer.

[0052] The risk identification module is used to perform dynamic pattern recognition on the initial signal set of multi-point leakage current using a convolutional neural network, extract feature vectors, and make risk judgments based on the feature vectors in order to identify potential leakage current risk nodes.

[0053] The current shunting decision module is used to obtain auxiliary channel configuration parameters from the identified potential leakage risk nodes, calculate and determine one or more target auxiliary channels for shunting leakage current in real time, and generate corresponding channel switching instructions.

[0054] The channel execution module is used to activate the target auxiliary channel in response to the channel switching command to establish a leakage current shunt path;

[0055] The impedance adjustment module is used to dynamically adjust the impedance of the target auxiliary channel so that the distribution of the shunt current meets the load transfer stability requirements and completes uninterrupted circuit switching.

[0056] The technical solution of the present invention has the following advantages compared with the prior art:

[0057] This invention provides a leakage current protection method based on bypass cabin circuit switching. In principle, it no longer relies on isolated, static passive protection, but instead constructs a closed-loop system of active perception, intelligent decision-making and dynamic control.

[0058] In principle, this method first constructs a comprehensive system status perception network by continuously collecting electrical data from multiple nodes in real time. This fundamentally solves the blind spot problem of traditional single-point monitoring and provides a data foundation for identifying dynamically changing leakage risks. Subsequently, a convolutional neural network is introduced for dynamic pattern recognition. This is fundamentally different from the traditional mechanism that relies on fixed thresholds to trigger alarms. The convolutional neural network can learn and extract deep feature patterns that characterize abnormal current transfer from massive, seemingly chaotic multi-point signals, thereby achieving intelligent prediction of potential leakage risk nodes and greatly improving the timeliness and accuracy of risk detection.

[0059] After identifying the risk nodes, the solution did not adopt the crude method of cutting off the circuit, but instead used a diversion strategy. Based on the information of the risk nodes, one or more pre-prepared target auxiliary channels were calculated and activated in real time. By dynamically adjusting the impedance of these diversion channels, the magnitude and distribution of the shunt current can be precisely controlled to ensure that most of the current is safely guided to a harmless path. At the same time, the switching process of the main circuit is not affected in any way, and the power supply remains stable.

[0060] In summary, the beneficial effects of this technical solution include the following three points: First, through intelligent pattern recognition using convolutional neural networks, early and accurate warnings of dynamic, multi-point leakage risks are achieved, preventing accidents from escalating. Second, by calculating and activating auxiliary shunt channels in real time, active diversion of leakage current is achieved instead of passive disconnection, fundamentally resolving the contradiction between safety protection and power supply continuity, and achieving truly uninterrupted safe switching. Finally, this systematic approach greatly improves the safety and reliability of bypass shelter circuit switching operations, providing a higher level of protection for operators and power equipment. Attached Figure Description

[0061] To make the content of this invention easier to understand, the invention will be further described in detail below with reference to specific embodiments and accompanying drawings, wherein:

[0062] Figure 1 This is a flowchart of the leakage current protection method based on bypass cabin circuit switching of the present invention;

[0063] Figure 2 This is a flowchart of the steps for real-time continuous acquisition of voltage and current data of multiple nodes during the switching process of the bypass cabin circuit in this invention.

[0064] Figure 3This is a flowchart illustrating the steps of the present invention to perform dynamic pattern recognition and extract feature vectors from a set of initial signals of multi-point leakage current using a convolutional neural network;

[0065] Figure 4 This is a flowchart of the steps in this invention to identify potential leakage risk nodes based on feature vectors.

[0066] Figure 5 This is a flowchart of the steps for determining the target auxiliary channel in this invention;

[0067] Figure 6 This is a flowchart illustrating the steps of one embodiment of the present invention for determining a target auxiliary channel;

[0068] Figure 7 This is a flowchart illustrating the steps of another embodiment of the present invention for determining the target auxiliary channel;

[0069] Figure 8 This is a structural framework diagram of the leakage current protection system based on bypass cabin circuit switching of the present invention. Detailed Implementation

[0070] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, so that those skilled in the art can better understand and implement the present invention. However, the embodiments described are not intended to limit the present invention.

[0071] In existing technologies, circuit switching operations in power supply systems need to complete load transfer without interrupting power supply, which places higher demands on leakage current protection. Traditional methods rely on upgrading insulation levels or installing leakage current devices (RCDs). The former's protection capability decreases with equipment aging or in complex environments, while the latter can only cut off the circuit after a leakage occurs, failing to meet the continuous power supply requirements during dynamic switching. In temporary power supply scenarios such as bypass shelters, the circuit state changes rapidly during load transfer, and leakage risk points may appear at multiple nodes simultaneously. Traditional technologies cannot locate the leakage current source in real time, nor can they implement effective protection while maintaining power supply, resulting in both operator safety risks and potential equipment damage.

[0072] To address the aforementioned issues, this invention proposes a mechanism for rapid risk node location by combining real-time acquisition with dynamic pattern recognition, based on the study of multi-point signal correlation patterns. Further analysis reveals that simply cutting off the circuit cannot meet the requirements for power supply continuity; it is necessary to maintain system stability by actively diverting leakage current. Based on this, a leakage current guiding path based on an auxiliary channel is constructed, and current distribution is optimized through dynamic impedance adjustment, forming a complete uninterrupted protection closed loop.

[0073] Therefore, refer to Figure 1As shown, this application proposes a leakage current protection method including the following steps: real-time continuous acquisition of voltage and current data of multiple nodes during the switching process of the bypass cabin circuit to form a multi-point leakage current initial signal set during load transfer; dynamic pattern recognition of the multi-point leakage current initial signal set using a convolutional neural network (CNN), extracting feature vectors and performing risk assessment to identify potential leakage current risk nodes; obtaining auxiliary channel configuration parameters, calculating and determining the target auxiliary channel in real time and generating channel switching instructions; activating the target auxiliary channel to establish a leakage current shunt path; and dynamically adjusting the impedance of the target auxiliary channel to ensure that the shunt current distribution meets the load transfer stability requirements.

[0074] Among them, real-time continuous acquisition refers to the synchronous acquisition of electrical parameters of multiple nodes through a distributed sensor network. Specifically, it can be implemented using multiple sets of voltage-current sensing units, covering bus interfaces, load access points, and key branch nodes to form a complete circuit status monitoring system. Dynamic pattern recognition refers to the analysis of the spatiotemporal characteristics of signals through deep learning models. Specifically, it can be implemented using convolutional neural networks to perform local feature scanning and multi-level fusion on two-dimensional signal spectra to accurately capture abnormal signal correlation patterns. Auxiliary channel configuration parameters refer to the pre-set attribute parameters of backup conductive paths, which can include channel impedance values, current carrying capacity thresholds, and topological connection relationships, providing a calculation basis for current shunting decisions. Impedance dynamic adjustment refers to adjusting the impedance of conductive paths according to the deviation between the real-time current shunting current and the target range. Specifically, it can be implemented using controllable variable resistors, and closed-loop control is used to maintain stable system operation.

[0075] Specifically, during circuit switching startup, the distributed sensor network synchronously collects voltage and current signals from each node using a unified clock. After denoising, standardized data packets are generated. These data are reconstructed into a two-dimensional signal spectrum and input into a convolutional neural network. The network uses multi-layer convolutional kernels to scan and extract the correlation features of the signals in the temporal and spatial dimensions, forming a feature vector representing the current system state. The state assessment model determines whether there is a risk of leakage current based on the feature vector. When an anomaly is detected, the system searches a pre-set channel topology database and, combined with real-time load conditions, selects auxiliary channels that can be safely put into operation. After selecting a channel, the actuator immediately establishes a shunt path and dynamically adjusts the channel impedance by monitoring the actual shunt current to ensure that the current distribution remains within a stable range during load transfer.

[0076] Through the above technical solution, this application can identify multiple leakage risks in real time during load transfer, quickly establish the optimal shunt path, and maintain stable system operation through dynamic impedance adjustment. This solves the problem that traditional methods cannot simultaneously ensure power supply continuity and leakage protection, avoids power outages caused by leakage protection activation, reduces the risk of electric shock to operators and the probability of equipment damage, and ensures the safety and reliability of circuit switching operations.

[0077] Reference Figure 2 As shown, in order to implement the above method, this application further proposes a method for real-time continuous acquisition of voltage and current data of multiple nodes during the switching process of the bypass cabin circuit, including the following steps: During the switching process of the bypass cabin circuit, the voltage and current data of the bus interface, load access point and key branch nodes are acquired through a distributed sensor network. All sensors start measurement at a uniform sampling rate under the synchronization command. The differential amplifier circuit built into each sensor can suppress common-mode interference in the line, and the hardware filter eliminates high-frequency noise and power frequency harmonics. The digital signal after analog-to-digital conversion is appended with a timestamp accurate to the microsecond level and a unique node code to form a standardized data packet containing spatiotemporal information. After the data packet is transmitted to the central processing unit, the scattered local signals are reconstructed into a global signal set covering the entire circuit architecture by timestamp alignment and node code matching, providing a complete data foundation for subsequent leakage current analysis.

[0078] Among them, the voltage-current sensing unit refers to the sensor device deployed at key nodes of the circuit, which can be implemented by combining Hall effect sensors with differential amplifier circuits to capture voltage fluctuations and current changes in the nodes in real time; the synchronous acquisition command refers to the control signal that controls all sensors to start sampling synchronously, which can be implemented by a clock synchronization mechanism based on the IEEE 1588 protocol to ensure the timing consistency of data collected by different nodes; differential amplification and hardware filtering refers to the technology of preprocessing the raw sensor signals, which can be implemented by using an instrumentation amplifier with a bandpass filter to eliminate common-mode interference and high-frequency noise; timestamp and node identifier refers to the marking information attached to the data packet, which can be implemented by using a GPS timing module and preset node encoding rules for subsequent data alignment and traceability; alignment and reassembly refers to the operation of integrating multi-source data, which can be implemented by using a sliding window algorithm and data frame matching technology to construct a complete circuit state dataset.

[0079] Traditional methods rely on single-point detection or asynchronous acquisition, resulting in the loss of temporal correlation and spatial distribution characteristics of leakage current signals. This solution, through a distributed sensor network and synchronous acquisition mechanism, fully captures the dynamic electrical parameter changes of each node during circuit switching. Combined with hardware-level signal preprocessing and standardized data encapsulation, it effectively solves the problems of temporal inaccuracy and noise interference from multi-source heterogeneous data, providing reliable data support for accurately identifying multi-point leakage current risks.

[0080] Reference Figure 3As shown, this application further proposes a method for dynamic pattern recognition and feature vector extraction of the initial signal set of multi-point leakage current using a convolutional neural network. During the switching process of the bypass cabin circuit, the spatiotemporal distribution characteristics of the multi-point leakage current signal will dynamically change with the load transfer process. By dividing the original signal stream into a two-dimensional signal spectrum of equal length, the convolutional kernel can perform feature scanning along the two dimensions of time axis and node axis, effectively capturing the correlation between signal mutation points and adjacent nodes. The front-end convolutional layer of the hierarchical perception network identifies local waveform distortion features, and the deep network discovers abnormal propagation paths between multiple nodes through cross-channel feature fusion. The final comprehensive feature vector not only contains the leakage current intensity information at the current moment, but also implies the spatiotemporal correlation features of risk diffusion trend, providing high-dimensional feature support for subsequent risk assessment.

[0081] Standardized segmentation refers to dividing continuously acquired multi-node signals into equal-length time segments with a fixed duration. This can be achieved using a sliding window segmentation algorithm, eliminating phase differences caused by asynchronous sampling through a unified time reference. Hierarchical perception structure refers to a deep neural network formed by alternating stacks of multiple convolutional and pooling layers. This can be achieved using a residual connection structure, capturing spatiotemporal correlation patterns at different scales through progressively expanding receptive fields. Multi-level fusion refers to nonlinearly combining shallow local features with deep global features. This can be achieved using an attention mechanism, enhancing the saliency of key features through weighted aggregation. Comprehensive feature vector refers to reducing multidimensional features to a fixed length through fully connected layers. This can be achieved using a feature compression algorithm, forming a unique identifier representing the current dynamic state of the system through information condensation.

[0082] This scheme combines temporal and spatial correlation analysis by constructing a two-dimensional signal map. Utilizing the unique feature extraction capabilities of convolutional neural networks, it reveals the electrical coupling relationships between nodes while maintaining temporal continuity, significantly improving the accuracy of leakage current feature identification during dynamic switching. This application can extract physically meaningful spatiotemporal feature combinations from complex multi-node signals, accurately capturing leakage current risk features caused by factors such as poor contact and insulation degradation during load transfer. It solves the problem of misjudgment and omission caused by traditional methods neglecting the dynamic correlation between nodes, providing reliable input features for the subsequent precise location of risk nodes.

[0083] Furthermore, after extracting the feature vectors, risk assessment needs to be performed based on the feature vectors to identify potential leakage risk nodes, referring to... Figure 4As shown, during circuit switching, the multi-node signals collected are processed to form a comprehensive feature vector, which is then input into a pre-trained state assessment model. This model compares the current feature vector with the normal patterns and known leakage patterns in historical data, and outputs a three-dimensional assessment result including a risk level score, risk type identifier, and a list of key nodes. When the risk level score exceeds the warning threshold dynamically adjusted according to the current load rate, the system triggers a risk confirmation process. Combining the fault mechanism corresponding to the risk type identifier with the weight ranking in the list of key nodes, the system quickly locates the circuit node that contributes the most to the abnormal signal as a potential leakage risk point. For example, when an insulation degradation risk is detected, the system prioritizes checking nodes that are easily affected by the environment, such as load connection points. When a ground fault is determined, the system focuses on checking the connection status of high-current paths such as bus interfaces.

[0084] The state assessment model refers to a leakage risk classifier built using machine learning algorithms, specifically support vector machines or deep neural networks. It undergoes supervised training by inputting multi-dimensional feature vectors from historical normal operating conditions and preset leakage scenarios, enabling the model to identify abnormal signals during dynamic switching. The three-dimensional risk assessment result is a composite judgment index containing risk intensity, fault type, and key node information. It can be generated using a multi-task learning framework. The risk level score reflects the severity of leakage, the risk type identifier distinguishes different scenarios such as grounding faults or insulation degradation, and the list of nodes with the highest risk contribution is obtained by ranking the contribution weights of each node to the abnormal features using a backpropagation algorithm. The dynamic risk strategy refers to a warning threshold mechanism adjusted according to real-time load status, such as automatically lowering the risk level trigger threshold when the load rate exceeds a set value, to adapt to the safety protection requirements under different operating conditions.

[0085] This application can identify the location and fault type of potential leakage risk nodes in real time during circuit switching, significantly shortening the fault location time. By dynamically adjusting the risk judgment threshold, the system can avoid misjudgment caused by load fluctuations, ensuring power supply continuity while improving the accuracy of safety protection. The risk contribution ranking mechanism helps maintenance personnel quickly locate key fault points, reduce unnecessary node troubleshooting work, and improve emergency response efficiency.

[0086] Reference Figure 5As shown, this application further proposes specific steps for obtaining auxiliary channel configuration parameters from identified potential leakage risk nodes, and for calculating and determining one or more target auxiliary channels for shunting leakage current in real time: When a potential leakage risk node is detected, all available backup channels are first retrieved from a pre-set topology database based on its location to form an initial candidate set. Then, by combining the estimated leakage current intensity with the real-time load data of the channels, overload risk channels are eliminated through dynamic capacity margin calculation, and qualified channels are retained. Furthermore, the system stability after each qualified channel is put into operation is verified through power flow simulation to eliminate channels that may cause voltage drops or equipment overloads. Finally, the path with the lowest equivalent impedance is selected from the remaining safe channels as the target channel to ensure that the conduction efficiency of the shunting path is maximized.

[0087] Among them, channel topology refers to the pre-established electrical connection structure between backup auxiliary channels and main circuit nodes, which can be stored in the form of graph database or adjacency matrix for quick retrieval of backup channels electrically connected to risk nodes; current carrying capacity matching calculation refers to determining whether a channel has the ability to carry the estimated leakage current based on its rated current carrying capacity and current load rate, which can be implemented using dynamic capacity margin algorithm to screen out channels that meet current carrying requirements; power flow distribution impact simulation refers to predicting the voltage and current changes of each node after channel commissioning by establishing a circuit equivalent model, which can be performed iteratively using the Newton-Raphson method to verify the system safety after channel commissioning.

[0088] This solution employs a multi-level screening mechanism to sequentially complete connectivity retrieval, current carrying capacity verification, and system stability prediction. It can accurately identify target channels that meet current diversion requirements while avoiding secondary problems, thus resolving the cascading failures caused by improper channel selection during dynamic switching. Through this technical solution, this application can rapidly complete the entire process from candidate channel selection to optimal path determination, effectively preventing secondary failures caused by channel overload or system instability, and ensuring the reliability of leakage current diversion and power supply continuity during bypass shelter circuit switching.

[0089] Specifically, in actual implementation, the following two problems were encountered: one is that the current shunting capacity of a single channel is insufficient to achieve leakage current shunting; the other is that when selecting a current shunting channel, only the channel's own current carrying capacity is usually considered, while the capacity limit of the upstream power supply is ignored, which may lead to system collapse due to leakage current shunting operation exceeding the power supply's carrying capacity. To solve the above two practical problems, this invention further optimizes the process of determining the target auxiliary channel.

[0090] Reference Figure 6As shown, this application further proposes an improved scheme for calculating and determining one or more target auxiliary channels for shunting leakage current in real time when there is no single channel in the safety channel subset that meets the shunting requirements. Specifically, when it is detected that a single channel cannot meet the leakage current shunting requirements, the system automatically traverses all possible parallel channel combinations to form a candidate scheme pool. For each candidate scheme, a circuit model including line impedance, load characteristics, and power supply capacity is established to simulate the current distribution under different shunting ratios. During the model calculation process, power quality indicators such as voltage distortion rate and frequency deviation of each scheme are evaluated simultaneously, as well as thermal stability parameters such as transformer winding temperature rise and circuit breaker contact temperature. Through comprehensive comparison of multi-dimensional indicators, the scheme with the least impact on power quality and the safest equipment operation is selected as the final execution target. After the scheme is determined, the target channels are activated sequentially and the shunting weight of each channel is dynamically adjusted according to the preset phased switching strategy to ensure that the continuity of load power supply is not affected during the switching process.

[0091] Among them, the parallel cooperative channel combination refers to a composite channel structure formed by connecting two or more electrically connected backup auxiliary channels in parallel. Specifically, it can be implemented by using a topology analysis algorithm to traverse all suitable channel connection methods. Its function is to improve the overall current shunting capacity through multi-channel cooperative operation. The cooperative current shunting model is a mathematical model describing the current distribution relationship between parallel channels and the system operating state. Specifically, it can be constructed using circuit equations based on Kirchhoff's laws combined with power flow calculation algorithms, used to quantitatively evaluate the impact of different current shunting ratios on system stability. The overall power quality index refers to comprehensive parameters reflecting voltage fluctuations, harmonic content, and three-phase balance. Specifically, it can be normalized using a weighted scoring method for each power parameter. The system is divided into three parts: processing, measurement, and control. The processing measures the impact of the current sharing scheme on power quality. Thermal stability margin refers to the difference between the temperature rise of electrical equipment under current sharing conditions and the allowable temperature rise limit. This margin can be calculated using thermodynamic simulation combined with real-time temperature monitoring data to assess the long-term operational safety of the current sharing scheme. The optimal coordinated current sharing scheme is a balanced scheme that simultaneously satisfies optimal power quality and maximum thermal stability margin. This can be achieved by using a multi-objective optimization algorithm to rank and filter candidate schemes, ensuring optimal overall system performance under complex operating conditions. The coordinated switching instruction set is a set of control parameters including channel activation sequence, current sharing ratio, and impedance preset values. This set can be generated by a timing logic controller to achieve coordinated switching and precise control of multiple channels.

[0092] This solution, through the combined application and coordinated control of parallel channels, not only expands the upper limit of current shunting capacity, but also achieves a balance between system safety and power quality through optimized algorithms, avoiding the risk of secondary faults caused by overcurrent or overheating.

[0093] Reference Figure 7As shown, this application further proposes a system-level conflict verification process in the step of determining the target auxiliary channel from the subset of safe channels: When determining the target auxiliary channel, the total current load power of the system is first obtained through the sensor network, and the remaining capacity data of the upstream power supply corresponding to each candidate channel is extracted from the power management unit; the estimated leakage intensity of potential leakage nodes is converted into shunt power demand, which is then added to the original load of the candidate channel to form the total demand power; by comparing the total demand power with the remaining power supply capacity in real time, a set of candidate channels with sufficient capacity is selected. For example, when the remaining capacity of the upstream power supply of a candidate channel is 1000W, and the total demand power reaches 1200W after the channel is put into operation, the channel will be excluded, and only the channel that passes the capacity verification will be selected as the target auxiliary channel to ensure that the shunt operation will not cause power overload.

[0094] System-level conflict verification refers to the process of dynamically verifying the capacity of the upstream power supply for candidate auxiliary channels. This can be achieved by using a monitoring module deployed at the power supply output to collect remaining capacity data in real time. This step is used to avoid power supply overload due to shunting operations. The remaining capacity of the upstream power supply refers to the unused output power margin under current operating conditions. This can be calculated using a capacity calculation algorithm based on real-time current monitoring. This parameter determines the upper limit of the shunting capacity of the candidate channel. Total required power refers to the sum of the leakage current shunting power required after the candidate channel is put into operation and the original load power. This can be calculated using a power superposition model. This parameter is used to evaluate whether the power supply capacity meets the shunting requirements.

[0095] This solution introduces a system-level conflict verification mechanism to predict power capacity during the channel selection phase, effectively avoiding cascading failures caused by insufficient capacity. Through the above technical solution, this application can prevent system crashes caused by leakage current shunting operations exceeding the power supply's carrying capacity, ensuring that the power supply remains within a safe operating range during bypass cabin circuit switching, while maintaining power continuity during load transfer.

[0096] During the bypass module circuit switching process, once the target auxiliary channel is activated, its impedance value needs to be dynamically optimized based on real-time operating conditions. The actual shunt current flowing through the target auxiliary channel is monitored in real time. The actual shunt current value is compared with the target current range calculated based on load transfer stability requirements to generate a current deviation signal. For example, in the initial stage of load transfer, if the actual shunt current exceeds the preset upper limit, it indicates that the current channel impedance is too low, leading to excessive shunt current. In this case, increasing the impedance value can limit current growth. Conversely, if the current is below the lower limit, the shunt capacity is increased by decreasing the impedance. Furthermore, during impedance adjustment, the current deviation signal is analyzed for trend changes to predict its next direction and magnitude. Combining the current actual shunt current value with its predicted information, a comprehensive set of adjustment parameters is retrieved from a preset adjustment strategy mapping table. This set defines the intensity and rate of the adjustment. Based on the adjustment parameter set, a smooth, phased impedance adjustment command sequence is generated to replace a single, immediate adjustment command.

[0097] Real-time monitoring of the actual shunt current value refers to continuously acquiring real-time current data of the target auxiliary channel through a current sensor, specifically using a Hall effect sensor or Rogowski coil, to accurately capture dynamic current changes in the shunt path. The target current range refers to the safe operating range pre-calculated based on system stability requirements during load transfer, specifically determined through circuit simulation or historical operating data modeling, used to establish a control benchmark for the shunt current. The current deviation signal is the quantified result of the difference between the actual measured value and the target range, specifically generated using a proportional-integral-differential algorithm, used to characterize the degree of deviation between the current shunt state and the ideal state. Trend analysis refers to time-series prediction of the current deviation signal, specifically implemented using a sliding window algorithm combined with a linear regression model, used to predict the current change trend within the next few milliseconds. The adjustment strategy mapping table is a table storing the correspondence between different deviation signal characteristics and adjustment parameters, specifically implemented using a database or hash table structure, used for quickly matching the optimal adjustment strategy. The phased impedance adjustment command sequence refers to a set of commands that decompose a single adjustment into multiple progressive steps, specifically implemented using a multi-level command queue marked with timestamps, used to achieve a smooth transition of impedance values.

[0098] Through the above technical solution, this application can achieve a smooth transition in the impedance adjustment process, avoid current surges caused by sudden adjustments, ensure that the bypass container maintains power supply stability during load transfer, and effectively coordinate the contradiction between adjustment speed and system stability by executing the phased command sequence. It can quickly eliminate current deviations while maintaining the continuous adjustability of circuit parameters.

[0099] Reference Figure 8As shown, in order to implement the above-mentioned leakage current protection method, the present invention further discloses a leakage current protection system based on bypass cabin circuit switching, which is used to implement the above-mentioned leakage current protection method, including:

[0100] The data acquisition module is used to continuously acquire voltage and current data of multiple nodes during the switching process of the bypass cabin circuit in real time, forming a set of initial leakage signals at multiple points during load transfer.

[0101] The risk identification module is used to perform dynamic pattern recognition on the initial signal set of multi-point leakage current using a convolutional neural network, extract feature vectors, and make risk judgments based on the feature vectors in order to identify potential leakage current risk nodes.

[0102] The current shunting decision module is used to obtain auxiliary channel configuration parameters from the identified potential leakage risk nodes, calculate and determine one or more target auxiliary channels for shunting leakage current in real time, and generate corresponding channel switching instructions.

[0103] The channel execution module is used to activate the target auxiliary channel in response to the channel switching command to establish a leakage current shunt path;

[0104] The impedance adjustment module is used to dynamically adjust the impedance of the target auxiliary channel so that the distribution of the shunt current meets the load transfer stability requirements and completes uninterrupted circuit switching.

[0105] The leakage current protection system of this invention achieves millisecond-level location of risk nodes through multi-node synchronous monitoring and dynamic pattern recognition. Combined with intelligent current diversion strategy, it eliminates leakage risks while maintaining power supply continuity. It realizes real-time perception and active protection of leakage risks during bypass cabin circuit switching and can complete load transfer operations without power interruption.

[0106] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A leakage current protection method based on bypass cabin circuit switching, characterized in that, Includes the following steps: Real-time continuous acquisition of voltage and current data of multiple nodes during the switching process of the bypass cabin circuit constitutes a set of initial leakage signals at multiple points during load transfer. A convolutional neural network is used to perform dynamic pattern recognition on the initial signal set of multiple leakage currents, extract feature vectors, and make risk judgments based on the feature vectors to identify potential leakage current risk nodes. The process involves: obtaining auxiliary channel configuration parameters from identified potential leakage current risk nodes; calculating and determining one or more target auxiliary channels for shunting leakage current in real time; and generating corresponding channel switching instructions. Specifically, this includes: retrieving all electrically connected backup auxiliary channels from a pre-defined channel topology based on the location information of potential leakage current risk nodes, forming an initial candidate channel set; performing current carrying capacity matching calculations based on the estimated leakage current intensity of potential leakage current risk nodes and the real-time load capacity of each candidate channel, screening out all channels with qualified carrying capacity, forming a qualified channel subset; simulating the impact of each channel in the qualified channel subset on the power flow distribution of the entire circuit system after its activation, excluding channels that would cause voltage overruns or equipment overloads at any node, forming a safe channel subset; and selecting the channel with the shortest path and lowest equivalent impedance from the safe channel subset as the target auxiliary channel. When there is no single channel in the safe channel subset that meets the current shunting requirements, one or more target auxiliary channels for shunting leakage current are calculated and determined in real time. This further includes: generating all possible parallel cooperative channel combinations consisting of two or more channels based on the qualified channel subset; constructing a cooperative current shunting model for each parallel cooperative channel combination, and calculating the overall power quality index and thermal stability margin of the system when the leakage current is distributed among the channels in a specific ratio; selecting the combination that optimizes the overall power quality index and maximizes the thermal stability margin from all feasible cooperative channel combinations as the optimal cooperative current shunting scheme; and generating a cooperative switching instruction set containing the input timing of each target auxiliary channel and its preset current shunting weights based on the optimal cooperative current shunting scheme as the final channel switching instruction. The step of determining the target auxiliary channel from the subset of safe channels further includes performing system-level conflict verification: obtaining the total power of all loads currently in operation in the system, and the remaining capacity of the upstream power supply for each candidate auxiliary channel; superimposing the estimated leakage current intensity of potential leakage risk nodes with the additional shunt power required by the candidate channel to calculate the total power demand after the channel is put into operation; comparing the total power demand with the remaining capacity of the corresponding upstream power supply to verify whether the power supply capacity is sufficient; and finally determining the target auxiliary channel from the candidate channels that pass the verification. In response to a channel switching command, the target auxiliary channel is activated to establish a leakage current shunt path; The impedance of the target auxiliary channel is dynamically adjusted to ensure that the distribution of the shunt current meets the load transfer stability requirements, thus completing uninterrupted circuit switching.

2. The leakage current protection method based on bypass cabin circuit switching according to claim 1, characterized in that: Real-time continuous acquisition of voltage and current data from multiple nodes during the switching process of the bypass shelter circuit constitutes a multi-point leakage current initial signal set during load transfer, specifically including: Multiple voltage-current sensing units are deployed at the bus interface, load access point, and key switching branch nodes of the bypass cabin circuit to form a distributed sensing network covering the circuit architecture. Broadcast a unified synchronization acquisition command to the distributed sensor network to coordinate all voltage-current sensing units to start data acquisition with the same sampling clock; Inside each voltage-current sensing unit, the acquired raw analog signal is differentially amplified and hardware filtered to obtain the denoised local signal. The local signals of each sensing unit after denoising are converted from analog to digital and stamped with a unified timestamp and node identifier to form a standardized node data packet. All node data packets are aligned and reassembled based on timestamps and node identifiers to generate a multi-point leakage initial signal set.

3. The leakage current protection method based on bypass cabin circuit switching according to claim 1, characterized in that: Dynamic pattern recognition of multi-point leakage current initial signal sets using convolutional neural networks to extract feature vectors, specifically including: The initial signals of multi-point leakage current are standardized and segmented in the time dimension to construct a two-dimensional signal spectrum with time as the horizontal axis and multi-node signals as the vertical axis. Construct a feature extraction network with a hierarchical perception structure. The front-end structure of this network is used to perform local feature scanning on the two-dimensional signal spectrum and capture the local correlation patterns of the signal in the time and node dimensions. By utilizing the deep structure of the feature extraction network, multi-level fusion and abstraction are performed on the local correlation patterns obtained from scanning to form a high-order temporal feature combination; The high-order temporal features are combined and mapped into a fixed-dimensional comprehensive feature vector that condenses the dynamic switching state information of the current time period.

4. The leakage current protection method based on bypass cabin circuit switching according to claim 3, characterized in that: Risk assessment is performed based on feature vectors to identify potential leakage risk nodes, specifically including: The comprehensive feature vector is input into a pre-trained state assessment model, which is trained based on feature vector data from historical normal operation and known leakage states; The status assessment model outputs a three-dimensional risk assessment result, which includes a risk level score, a risk type identifier, and a list of nodes with the highest risk contribution. Based on the preset dynamic risk strategy, the risk assessment results are analyzed. If the risk level score exceeds the dynamic warning line under the current system load, risk confirmation is triggered. After risk confirmation is triggered, one or more potential leakage risk nodes are identified by combining the risk type identifier with the list of nodes with the highest risk contribution.

5. The leakage current protection method based on bypass cabin circuit switching according to claim 1, characterized in that: Dynamically adjust the impedance of the target auxiliary channel, specifically including: Real-time monitoring of the actual shunt current value flowing through the target auxiliary channel; The actual shunt current value is compared with the target current range calculated based on the load transfer stability requirements to generate a current deviation signal; Based on the magnitude and direction of the current deviation signal, a corresponding impedance adjustment command is generated; if the actual value is higher than the upper limit of the target range, the command is to increase the impedance; if the actual value is lower than the lower limit of the target range, the command is to decrease the impedance.

6. The leakage current protection method based on bypass cabin circuit switching according to claim 5, characterized in that: Based on the magnitude and direction of the current deviation signal, a corresponding impedance adjustment command is generated, further including: Analyze the changing trend of the current deviation signal to predict its next direction and magnitude of change; Based on the current actual shunt current value and its prediction information, a comprehensive combination of adjustment parameters is retrieved from a preset adjustment strategy mapping table. This combination defines the intensity and rate of this adjustment. Based on the combination of adjustment parameters, a smooth, phased sequence of impedance adjustment commands is generated to replace a single instantaneous adjustment command.

7. A leakage current protection system based on bypass cabin circuit switching, used to implement the leakage current protection method according to any one of claims 1 to 6, characterized in that: include: The data acquisition module is used to continuously acquire voltage and current data of multiple nodes during the switching process of the bypass cabin circuit in real time, forming a set of initial leakage signals at multiple points during load transfer. The risk identification module is used to perform dynamic pattern recognition on the initial signal set of multi-point leakage current using a convolutional neural network, extract feature vectors, and make risk judgments based on the feature vectors in order to identify potential leakage current risk nodes. The current shunting decision module is used to obtain auxiliary channel configuration parameters from the identified potential leakage risk nodes, calculate and determine one or more target auxiliary channels for shunting leakage current in real time, and generate corresponding channel switching instructions. The channel execution module is used to activate the target auxiliary channel in response to the channel switching command to establish a leakage current shunt path; The impedance adjustment module is used to dynamically adjust the impedance of the target auxiliary channel so that the distribution of the shunt current meets the load transfer stability requirements and completes uninterrupted circuit switching.

Images